chore(graph): Remove obsolete linear nodes and legacy tests

Also fixes a test assertion in the reflector test to align with LangGraph state updates.

README.md

@@ -4,6 +4,7 @@ A stateful, graph-based chatbot for election data analysis, built with LangGraph
 
 ## 🚀 Features
 
+- **Multi-Agent Orchestration**: Decomposes complex queries and delegates them to specialized sub-agents (Data Analyst, Researcher) using a robust feedback loop.
 - **Intelligent Query Analysis**: Automatically determines if a query needs data analysis, web research, or clarification.
 - **Automated Data Analysis**: Generates and executes Python code to analyze election datasets and produce visualizations.
 - **Web Research**: Integrates web search capabilities for general election-related questions.

backend/.env.example

@@ -57,3 +57,11 @@ OIDC_REDIRECT_URI=http://localhost:8000/api/v1/auth/oidc/callback
 # Researcher
 # RESEARCHER_LLM__PROVIDER=google
 # RESEARCHER_LLM__MODEL=gemini-2.0-flash
+
+# Reflector
+# REFLECTOR_LLM__PROVIDER=openai
+# REFLECTOR_LLM__MODEL=gpt-5-mini
+
+# Synthesizer
+# SYNTHESIZER_LLM__PROVIDER=openai
+# SYNTHESIZER_LLM__MODEL=gpt-5-mini

backend/GEMINI.md

@@ -1,12 +1,13 @@
 # Election Analytics Chatbot - Backend Guide
 
 ## Overview
-The backend is a Python-based FastAPI application that leverages **LangGraph** to provide a stateful, agentic workflow for election data analysis. It handles complex queries by decomposing them into tasks such as data analysis, web research, or user clarification.
+The backend is a Python-based FastAPI application that leverages **LangGraph** to provide a stateful, hierarchical multi-agent workflow for election data analysis. It handles complex queries using an Orchestrator-Workers pattern, decomposing tasks and delegating them to specialized subgraphs (Data Analyst, Researcher) with built-in reflection and error recovery.
 
 ## 1. Architecture Overview
-- **Framework**: LangGraph for workflow orchestration and state management.
+- **Framework**: LangGraph for hierarchical workflow orchestration and state management.
 - **API**: FastAPI for providing REST and streaming (SSE) endpoints.
-- **State Management**: Persistent state using LangGraph's `StateGraph` with a PostgreSQL checkpointer.
+- **State Management**: Persistent state using LangGraph's `StateGraph` with a PostgreSQL checkpointer. Maintains global state (`AgentState`) and isolated worker states (`WorkerState`).
+- **Virtual File System (VFS)**: An in-memory abstraction passed between nodes to manage intermediate artifacts (scripts, CSVs, charts) without bloating the context window.
 - **Database**: PostgreSQL.
   - Application data: Uses `users` table for local and OIDC users (String IDs).
   - History: Persists chat history and artifacts.
@@ -14,23 +15,28 @@ The backend is a Python-based FastAPI application that leverages **LangGraph** t
 
 ## 2. Core Components
 
-### 2.1. The Graph State (`src/ea_chatbot/graph/state.py`)
-The state tracks the conversation context, plan, generated code, execution results, and artifacts.
+### 2.1. State Management (`src/ea_chatbot/graph/state.py` & `workers/*/state.py`)
+- **Global State**: Tracks the conversation context, the high-level task `checklist`, execution progress (`current_step`), and the VFS.
+- **Worker State**: Isolated snapshot for specialized subgraphs, tracking internal retry loops (`iterations`), worker-specific prompts, and raw results.
 
-### 2.2. Nodes (The Actors)
+### 2.2. The Orchestrator
 Located in `src/ea_chatbot/graph/nodes/`:
 
-- **`query_analyzer`**: Analyzes the user query to determine the intent and required data.
-- **`planner`**: Creates a step-by-step plan for data analysis.
-- **`coder`**: Generates Python code based on the plan and dataset metadata.
-- **`executor`**: Safely executes the generated code and captures outputs (dataframes, plots).
-- **`error_corrector`**: Fixes code if execution fails.
-- **`researcher`**: Performs web searches for general election information.
-- **`summarizer`**: Generates a natural language response based on the analysis results.
-- **`clarification`**: Asks the user for more information if the query is ambiguous.
+- **`query_analyzer`**: Analyzes the user query to determine the intent and required data. If ambiguous, routes to `clarification`.
+- **`planner`**: Decomposes the user request into a strategic `checklist` of sub-tasks assigned to specific workers.
+- **`delegate`**: The traffic controller. Routes the current task to the appropriate worker and enforces a strict retry budget to prevent infinite loops.
+- **`reflector`**: The quality control node. Evaluates a worker's summary against the sub-task requirements. Can trigger a retry if unsatisfied.
+- **`synthesizer`**: Aggregates all worker results into a final, cohesive response for the user.
+- **`clarification`**: Asks the user for more information if the query is critically ambiguous.
 
-### 2.3. The Workflow (Graph)
-The graph connects these nodes with conditional edges, allowing for iterative refinement and error correction.
+### 2.3. Specialized Workers (Sub-Graphs)
+Located in `src/ea_chatbot/graph/workers/`:
+
+- **`data_analyst`**: Generates Python/SQL code, executes it securely, and captures dataframes/plots. Contains an internal retry loop (`coder` -> `executor` -> error check -> `coder`).
+- **`researcher`**: Performs web searches for general election information and synthesizes factual findings.
+
+### 2.4. The Workflow
+The global graph connects the Orchestrator nodes, wrapping the Worker subgraphs as self-contained nodes with mapped inputs and outputs.
 
 ## 3. Key Modules
 

backend/src/ea_chatbot/api/routers/agent.py

@@ -56,7 +56,8 @@ async def stream_agent_events(
                 initial_state, 
                 config, 
                 version="v2",
-                checkpointer=checkpointer 
+                checkpointer=checkpointer,
+                subgraphs=True
             ):
                 kind = event.get("event")
                 name = event.get("name")
@@ -71,8 +72,8 @@ async def stream_agent_events(
                     "data": data
                 }
                 
-                # Buffer assistant chunks (summarizer and researcher might stream)
-                if kind == "on_chat_model_stream" and node_name in ["summarizer", "researcher", "clarification"]:
+                # Buffer assistant chunks (synthesizer and clarification might stream)
+                if kind == "on_chat_model_stream" and node_name in ["synthesizer", "clarification"]:
                     chunk = data.get("chunk", "")
                     # Use utility to safely extract text content from the chunk
                     chunk_data = convert_to_json_compatible(chunk)
@@ -83,7 +84,7 @@ async def stream_agent_events(
                         assistant_chunks.append(str(chunk_data))
                 
                 # Buffer and encode plots
-                if kind == "on_chain_end" and name == "executor":
+                if kind == "on_chain_end" and name == "data_analyst_worker":
                     output = data.get("output", {})
                     if isinstance(output, dict) and "plots" in output:
                         plots = output["plots"]
@@ -95,7 +96,7 @@ async def stream_agent_events(
                         output_event["data"]["encoded_plots"] = encoded_plots
 
                 # Collect final response from terminal nodes
-                if kind == "on_chain_end" and name in ["summarizer", "researcher", "clarification"]:
+                if kind == "on_chain_end" and name in ["synthesizer", "clarification"]:
                     output = data.get("output", {})
                     if isinstance(output, dict) and "messages" in output:
                         last_msg = output["messages"][-1]

backend/src/ea_chatbot/config.py

@@ -52,6 +52,8 @@ class Settings(BaseSettings):
     coder_llm: LLMConfig = Field(default_factory=lambda: LLMConfig(model="gpt-5-mini", temperature=0.0))
     summarizer_llm: LLMConfig = Field(default_factory=lambda: LLMConfig(model="gpt-5-mini", temperature=0.0))
     researcher_llm: LLMConfig = Field(default_factory=lambda: LLMConfig(model="gpt-5-mini", temperature=0.0))
+    reflector_llm: LLMConfig = Field(default_factory=lambda: LLMConfig(model="gpt-5-mini", temperature=0.0))
+    synthesizer_llm: LLMConfig = Field(default_factory=lambda: LLMConfig(model="gpt-5-mini", temperature=0.0))
     
     # Allow nested env vars like QUERY_ANALYZER_LLM__MODEL
     model_config = SettingsConfigDict(env_nested_delimiter='__', env_prefix='')

backend/src/ea_chatbot/graph/nodes/clarification.py

@@ -36,8 +36,7 @@ Please ask the user for the necessary details."""
         response = llm.invoke(messages)
         logger.info("[bold green]Clarification generated.[/bold green]")
         return {
-            "messages": [response],
-            "next_action": "end" # To indicate we are done for now
+            "messages": [response]
         }
     except Exception as e:
         logger.error(f"Failed to generate clarification: {str(e)}")

backend/src/ea_chatbot/graph/nodes/coder.py

@@ -1,47 +0,0 @@
-from ea_chatbot.graph.state import AgentState
-from ea_chatbot.config import Settings
-from ea_chatbot.utils.llm_factory import get_llm_model
-from ea_chatbot.utils import database_inspection
-from ea_chatbot.utils.logging import get_logger, LangChainLoggingHandler
-from ea_chatbot.graph.prompts.coder import CODE_GENERATOR_PROMPT
-from ea_chatbot.schemas import CodeGenerationResponse
-
-def coder_node(state: AgentState) -> dict:
-    """Generate Python code based on the plan and data summary."""
-    question = state["question"]
-    plan = state.get("plan", "")
-    code_output = state.get("code_output", "None")
-    
-    settings = Settings()
-    logger = get_logger("coder")
-    
-    logger.info("Generating Python code...")
-    
-    llm = get_llm_model(
-        settings.coder_llm,
-        callbacks=[LangChainLoggingHandler(logger=logger)]
-    )
-    structured_llm = llm.with_structured_output(CodeGenerationResponse)
-    
-    # Always provide data summary
-    database_description = database_inspection.get_data_summary(data_dir=settings.data_dir) or "No data available."
-    example_code = "" # Placeholder
-    
-    messages = CODE_GENERATOR_PROMPT.format_messages(
-        question=question,
-        plan=plan,
-        database_description=database_description,
-        code_exec_results=code_output,
-        example_code=example_code
-    )
-    
-    try:
-        response = structured_llm.invoke(messages)
-        logger.info("[bold green]Code generated.[/bold green]")
-        return {
-            "code": response.parsed_code,
-            "error": None # Clear previous errors on new code generation
-        }
-    except Exception as e:
-        logger.error(f"Failed to generate code: {str(e)}")
-        raise e

backend/src/ea_chatbot/graph/nodes/delegate.py

@@ -0,0 +1,28 @@
+from ea_chatbot.graph.state import AgentState
+from ea_chatbot.utils.logging import get_logger
+
+def delegate_node(state: AgentState) -> dict:
+    """Determine which worker subgraph to call next based on the checklist."""
+    checklist = state.get("checklist", [])
+    current_step = state.get("current_step", 0)
+    iterations = state.get("iterations", 0)
+    logger = get_logger("orchestrator:delegate")
+    
+    if not checklist or current_step >= len(checklist):
+        logger.info("Checklist complete or empty. Routing to summarizer.")
+        return {"next_action": "summarize"}
+    
+    # Enforce retry budget
+    if iterations >= 3:
+        logger.error(f"Max retries reached for task {current_step}. Routing to summary with failure.")
+        return {
+            "next_action": "summarize",
+            "iterations": 0, # Reset for next turn
+            "summary": f"Failed to complete task {current_step} after {iterations} attempts."
+        }
+    
+    task_info = checklist[current_step]
+    worker = task_info.get("worker", "data_analyst")
+    
+    logger.info(f"Delegating next task to worker: {worker} (Attempt {iterations + 1})")
+    return {"next_action": worker}

backend/src/ea_chatbot/graph/nodes/error_corrector.py

@@ -1,44 +0,0 @@
-from ea_chatbot.graph.state import AgentState
-from ea_chatbot.config import Settings
-from ea_chatbot.utils.llm_factory import get_llm_model
-from ea_chatbot.utils.logging import get_logger, LangChainLoggingHandler
-from ea_chatbot.graph.prompts.coder import ERROR_CORRECTOR_PROMPT
-from ea_chatbot.schemas import CodeGenerationResponse
-
-def error_corrector_node(state: AgentState) -> dict:
-    """Fix the code based on the execution error."""
-    code = state.get("code", "")
-    error = state.get("error", "Unknown error")
-    
-    settings = Settings()
-    logger = get_logger("error_corrector")
-    
-    logger.warning(f"[bold red]Execution error detected:[/bold red] {error[:100]}...")
-    logger.info("Attempting to correct the code...")
-    
-    # Reuse coder LLM config or add a new one. Using coder_llm for now.
-    llm = get_llm_model(
-        settings.coder_llm,
-        callbacks=[LangChainLoggingHandler(logger=logger)]
-    )
-    structured_llm = llm.with_structured_output(CodeGenerationResponse)
-    
-    messages = ERROR_CORRECTOR_PROMPT.format_messages(
-        code=code,
-        error=error
-    )
-    
-    try:
-        response = structured_llm.invoke(messages)
-        logger.info("[bold green]Correction generated.[/bold green]")
-        
-        current_iterations = state.get("iterations", 0)
-        
-        return {
-            "code": response.parsed_code,
-            "error": None, # Clear error after fix attempt
-            "iterations": current_iterations + 1
-        }
-    except Exception as e:
-        logger.error(f"Failed to correct code: {str(e)}")
-        raise e

backend/src/ea_chatbot/graph/nodes/planner.py

@@ -1,34 +1,32 @@
-import yaml
 from ea_chatbot.graph.state import AgentState
 from ea_chatbot.config import Settings
 from ea_chatbot.utils.llm_factory import get_llm_model
 from ea_chatbot.utils import helpers, database_inspection
 from ea_chatbot.utils.logging import get_logger, LangChainLoggingHandler
 from ea_chatbot.graph.prompts.planner import PLANNER_PROMPT
-from ea_chatbot.schemas import TaskPlanResponse
+from ea_chatbot.schemas import ChecklistResponse
 
 def planner_node(state: AgentState) -> dict:
-    """Generate a structured plan based on the query analysis."""
+    """Generate a high-level task checklist for the Orchestrator."""
     question = state["question"]
     history = state.get("messages", [])[-6:]
     summary = state.get("summary", "")
     
     settings = Settings()
-    logger = get_logger("planner")
+    logger = get_logger("orchestrator:planner")
     
-    logger.info("Generating task plan...")
+    logger.info("Generating high-level task checklist...")
     
     llm = get_llm_model(
         settings.planner_llm,
         callbacks=[LangChainLoggingHandler(logger=logger)]
     )
-    structured_llm = llm.with_structured_output(TaskPlanResponse)
+    structured_llm = llm.with_structured_output(ChecklistResponse)
     
     date_str = helpers.get_readable_date()
     
-    # Always provide data summary; LLM decides relevance.
+    # Data summary for context
     database_description = database_inspection.get_data_summary(data_dir=settings.data_dir) or "No data available."
-    example_plan = ""
     
     messages = PLANNER_PROMPT.format_messages(
         date=date_str,
@@ -36,16 +34,20 @@ def planner_node(state: AgentState) -> dict:
         history=history,
         summary=summary,
         database_description=database_description,
-        example_plan=example_plan
+        example_plan="Decompose into data_analyst and researcher tasks."
     )
     
-    # Generate the structured plan
     try:
-        response = structured_llm.invoke(messages)
-        # Convert the structured response back to YAML string for the state
-        plan_yaml = yaml.dump(response.model_dump(), sort_keys=False)
-        logger.info("[bold green]Plan generated successfully.[/bold green]")
-        return {"plan": plan_yaml}
+        response = ChecklistResponse.model_validate(structured_llm.invoke(messages))
+        # Convert ChecklistTask objects to dicts for state
+        checklist = [task.model_dump() for task in response.checklist]
+        logger.info(f"[bold green]Checklist generated with {len(checklist)} tasks.[/bold green]")
+        return {
+            "checklist": checklist,
+            "current_step": 0,
+            "iterations": 0, # Reset iteration counter for the new plan
+            "summary": response.reflection # Use reflection as initial summary
+        }
     except Exception as e:
-        logger.error(f"Failed to generate plan: {str(e)}")
+        logger.error(f"Failed to generate checklist: {str(e)}")
         raise e

backend/src/ea_chatbot/graph/nodes/query_analyzer.py

@@ -1,18 +1,9 @@
-from typing import List, Literal
-from pydantic import BaseModel, Field
 from ea_chatbot.graph.state import AgentState
 from ea_chatbot.config import Settings
 from ea_chatbot.utils.llm_factory import get_llm_model
 from ea_chatbot.utils.logging import get_logger, LangChainLoggingHandler
 from ea_chatbot.graph.prompts.query_analyzer import QUERY_ANALYZER_PROMPT
-
-class QueryAnalysis(BaseModel):
-    """Analysis of the user's query."""
-    data_required: List[str] = Field(description="List of data points or entities mentioned (e.g., ['2024 results', 'Florida']).")
-    unknowns: List[str] = Field(description="List of target information the user wants to know or needed for final answer (e.g., 'who won', 'total votes').")
-    ambiguities: List[str] = Field(description="List of CRITICAL missing details that prevent ANY analysis. Do NOT include database names or plot types if defaults can be used.")
-    conditions: List[str] = Field(description="List of any filters or constraints (e.g., ['year=2024', 'state=Florida']). Include context resolved from history.")
-    next_action: Literal["plan", "clarify", "research"] = Field(description="The next action to take. 'plan' for data analysis (even with defaults), 'research' for general knowledge, or 'clarify' ONLY for critical ambiguities.")
+from ea_chatbot.schemas import QueryAnalysis
 
 def query_analyzer_node(state: AgentState) -> dict:
     """Analyze the user's question and determine the next course of action."""

backend/src/ea_chatbot/graph/nodes/reflector.py

@@ -0,0 +1,63 @@
+from ea_chatbot.graph.state import AgentState
+from ea_chatbot.config import Settings
+from ea_chatbot.utils.llm_factory import get_llm_model
+from ea_chatbot.utils.logging import get_logger, LangChainLoggingHandler
+from ea_chatbot.schemas import ReflectorResponse
+
+def reflector_node(state: AgentState) -> dict:
+    """Evaluate if the worker's output satisfies the current sub-task."""
+    checklist = state.get("checklist", [])
+    current_step = state.get("current_step", 0)
+    summary = state.get("summary", "") # This contains the worker's summary
+    
+    if not checklist or current_step >= len(checklist):
+        return {"next_action": "summarize"}
+    
+    task_info = checklist[current_step]
+    task_desc = task_info.get("task", "")
+    
+    settings = Settings()
+    logger = get_logger("orchestrator:reflector")
+    
+    logger.info(f"Evaluating worker output for task: {task_desc[:50]}...")
+    
+    llm = get_llm_model(
+        settings.reflector_llm,
+        callbacks=[LangChainLoggingHandler(logger=logger)]
+    )
+    structured_llm = llm.with_structured_output(ReflectorResponse)
+    
+    prompt = f"""You are a Lead Orchestrator evaluating the work of a specialized sub-agent.
+
+**Sub-Task assigned:**
+{task_desc}
+
+**Worker's Result Summary:**
+{summary}
+
+Evaluate if the result is satisfactory and complete for this specific sub-task. 
+If there were major errors or the output is missing critical data requested in the sub-task, mark satisfied as False."""
+
+    try:
+        response = structured_llm.invoke(prompt)
+        if response.satisfied:
+            logger.info("[bold green]Sub-task satisfied.[/bold green] Advancing plan.")
+            return {
+                "current_step": current_step + 1,
+                "iterations": 0, # Reset for next task
+                "next_action": "delegate"
+            }
+        else:
+            logger.warning(f"[bold yellow]Sub-task NOT satisfied.[/bold yellow] Reason: {response.reasoning}")
+            # Do NOT advance the step. Increment iterations to track retries.
+            return {
+                "iterations": state.get("iterations", 0) + 1, 
+                "next_action": "delegate"
+            }
+    except Exception as e:
+        logger.error(f"Failed to reflect: {str(e)}")
+        # On error, increment iterations to avoid infinite loop if LLM is stuck
+        return {
+            "iterations": state.get("iterations", 0) + 1,
+            "next_action": "delegate"
+        }

backend/src/ea_chatbot/graph/nodes/summarizer.py

@@ -1,43 +0,0 @@
-from ea_chatbot.graph.state import AgentState
-from ea_chatbot.config import Settings
-from ea_chatbot.utils.llm_factory import get_llm_model
-from ea_chatbot.utils.logging import get_logger, LangChainLoggingHandler
-from ea_chatbot.graph.prompts.summarizer import SUMMARIZER_PROMPT
-
-def summarizer_node(state: AgentState) -> dict:
-    """Summarize the code execution results into a final answer."""
-    question = state["question"]
-    plan = state.get("plan", "")
-    code_output = state.get("code_output", "")
-    history = state.get("messages", [])[-6:]
-    summary = state.get("summary", "")
-    
-    settings = Settings()
-    logger = get_logger("summarizer")
-    
-    logger.info("Generating final summary...")
-    
-    llm = get_llm_model(
-        settings.summarizer_llm,
-        callbacks=[LangChainLoggingHandler(logger=logger)]
-    )
-    
-    messages = SUMMARIZER_PROMPT.format_messages(
-        question=question,
-        plan=plan,
-        code_output=code_output,
-        history=history,
-        summary=summary
-    )
-    
-    try:
-        response = llm.invoke(messages)
-        logger.info("[bold green]Summary generated.[/bold green]")
-        
-        # Return the final message to be added to the state
-        return {
-            "messages": [response]
-        }
-    except Exception as e:
-        logger.error(f"Failed to generate summary: {str(e)}")
-        raise e

backend/src/ea_chatbot/graph/nodes/synthesizer.py

@@ -0,0 +1,56 @@
+from ea_chatbot.graph.state import AgentState
+from ea_chatbot.config import Settings
+from ea_chatbot.utils.llm_factory import get_llm_model
+from ea_chatbot.utils.logging import get_logger, LangChainLoggingHandler
+from ea_chatbot.graph.prompts.synthesizer import SYNTHESIZER_PROMPT
+
+def synthesizer_node(state: AgentState) -> dict:
+    """Synthesize the results from multiple workers into a final answer."""
+    question = state["question"]
+    history = state.get("messages", [])
+    
+    # We look for the 'summary' from the last worker which might have cumulative info
+    # Or we can look at all messages in history bubbled up from workers.
+    # For now, let's assume the history contains all the worker summaries.
+    
+    settings = Settings()
+    logger = get_logger("orchestrator:synthesizer")
+    
+    logger.info("Synthesizing final answer from worker results...")
+    
+    llm = get_llm_model(
+        settings.synthesizer_llm,
+        callbacks=[LangChainLoggingHandler(logger=logger)]
+    )
+    
+    # Artifact summary
+    plots = state.get("plots", [])
+    vfs = state.get("vfs", {})
+    artifacts_summary = ""
+    if plots:
+        artifacts_summary += f"- {len(plots)} generated plot(s) are attached to this response.\n"
+    if vfs:
+        artifacts_summary += "- Data files available in VFS: " + ", ".join(vfs.keys()) + "\n"
+    if not artifacts_summary:
+        artifacts_summary = "No additional artifacts generated."
+
+    # We provide the full history and the original question
+    messages = SYNTHESIZER_PROMPT.format_messages(
+        question=question,
+        history=history,
+        worker_results="Review the worker summaries provided in the message history.",
+        artifacts_summary=artifacts_summary
+    )
+    
+    try:
+        response = llm.invoke(messages)
+        logger.info("[bold green]Final synthesis complete.[/bold green]")
+        
+        # Return the final message to be added to the state
+        return {
+            "messages": [response],
+            "next_action": "end"
+        }
+    except Exception as e:
+        logger.error(f"Failed to synthesize final answer: {str(e)}")
+        raise e

backend/src/ea_chatbot/graph/prompts/coder.py

@@ -9,6 +9,14 @@ The user will provide a task and a plan.
 - Do NOT assume a dataframe `df` is already loaded unless explicitly stated. You usually need to query it first.
 - The database schema is described in the prompt. Use it to construct valid SQL queries.
 
+**Virtual File System (VFS):**
+- An in-memory file system is available as `vfs`. Use it to persist intermediate data or large artifacts.
+- `vfs.write(filename, content, metadata=None)`: Save a file (content can be any serializable object).
+- `vfs.read(filename) -> (content, metadata)`: Read a file.
+- `vfs.list() -> list[str]`: List all files.
+- `vfs.delete(filename)`: Delete a file.
+- Prefer using VFS for intermediate DataFrames or complex data structures instead of printing everything.
+
 **Plotting:**
 - If you need to plot any data, use the `plots` list to store the figures.
 - Example: `plots.append(fig)` or `plots.append(plt.gcf())`.
@@ -18,7 +26,8 @@ The user will provide a task and a plan.
 - Produce FULL, COMPLETE CODE that includes all steps and solves the task!
 - Always include the import statements at the top of the code (e.g., `import pandas as pd`, `import matplotlib.pyplot as plt`).
 - Always include print statements to output the results of your code.
-- Use `db.query_df("SELECT ...")` to get data."""
+- Use `db.query_df("SELECT ...")` to get data.
+"""
 
 CODE_GENERATOR_USER = """TASK:
 {question}
@@ -43,6 +52,7 @@ Return a complete, corrected python code that incorporates the fixes for the err
 - You have access to a database client via the variable `db`.
 - Use `db.query_df(sql)` to run queries.
 - Use `plots.append(fig)` for plots.
+- You have access to `vfs` for persistent in-memory storage.
 - Always include imports and print statements."""
 
 ERROR_CORRECTOR_USER = """FAILED CODE:

backend/src/ea_chatbot/graph/prompts/planner.py

@@ -1,41 +1,29 @@
 from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
 
-PLANNER_SYSTEM = """You are a Research Specialist and your job is to find answers and educate the user. 
-Provide factual information responding directly to the user's question. Include key details and context to ensure your response comprehensively answers their query.
+PLANNER_SYSTEM = """You are a Lead Orchestrator for an Election Analytics Chatbot.
+Your job is to decompose complex user queries into a high-level checklist of tasks.
+
+**Specialized Workers:**
+1. `data_analyst`: Handles SQL queries, Python data analysis, and plotting. Use this when the user needs numbers, trends, or charts from the internal database.
+2. `researcher`: Performs web searches for current news, facts, or external data not in the primary database.
+
+**Orchestration Strategy:**
+- Analyze the user's question and the available data summary.
+- Create a logical sequence of tasks (checklist) for these workers.
+- Be specific in the task description for the worker (e.g., "Find the total votes in Florida 2020").
+- If the query is ambiguous, the Orchestrator loop will later handle clarification, but for now, make the best plan possible.
 
 Today's Date is: {date}"""
 
 PLANNER_USER = """Conversation Summary: {summary}
 
-TASK:
+USER QUESTION:
 {question}
 
-AVAILABLE DATA SUMMARY (Use only if relevant to the task):
+AVAILABLE DATABASE SUMMARY:
 {database_description}
 
-First: Evaluate whether you have all necessary and requested information to provide a solution. 
-Use the dataset description above to determine what data and in what format you have available to you.
-You are able to search internet if the user asks for it, or you require any information that you can not derive from the given dataset or the instruction.
-
-Second: Incorporate any additional relevant context, reasoning, or details from previous interactions or internal chain-of-thought that may impact the solution.
-Ensure that all such information is fully included in your response rather than referring to previous answers indirectly.
-
-Third: Reflect on the problem and briefly describe it, while addressing the problem goal, inputs, outputs,
-rules, constraints, and other relevant details that appear in the problem description.
-
-Fourth: Based on the preceding steps, formulate your response as an algorithm, breaking the solution in up to eight simple concise yet descriptive, clear English steps. 
-You MUST Include all values or instructions as described in the above task, or retrieved using internet search!
-If fewer steps suffice, that's acceptable. If more are needed, please include them.
-Remember to explain steps rather than write code.
-
-This algorithm will be later converted to Python code.
-If a dataframe is required, assume it is named 'df' and is already defined/populated based on the data summary.
-
-There is a list variable called `plots` that you need to use to store any plots you generate. Do not use `plt.show()` as it will render the plot and cause an error.
-
-Output the algorithm as a YAML string. Always enclose the YAML string within ```yaml tags.
-
-**Note: Ensure that any necessary context from prior interactions is fully embedded in the plan. Do not use phrases like "refer to previous answer"; instead, provide complete details inline.**
+Decompose the question into a strategic checklist. For each task, specify which worker should handle it.
 
 {example_plan}"""
 

backend/src/ea_chatbot/graph/prompts/synthesizer.py

@@ -0,0 +1,31 @@
+from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
+
+SYNTHESIZER_SYSTEM = """You are a Lead Orchestrator for an Election Analytics Chatbot.
+You have coordinated several specialized workers (Data Analysts, Researchers) to answer a user's complex query.
+
+Your goal is to synthesize their individual findings into a single, cohesive, and comprehensive final response for the user.
+
+**Guidelines:**
+- Do NOT mention the internal 'workers' or 'checklist' names.
+- Combine the data insights (from Data Analysts) and factual research (from Researchers) into a natural narrative.
+- Ensure all numbers, dates, and names from the worker reports are included accurately.
+- **Artifacts & Plots:** If plots or charts were generated, refer to them naturally (e.g., "The chart below shows..."). 
+- If any part of the plan failed, explain the status honestly but professionally.
+- Present data in clear formats (tables, bullet points) where appropriate."""
+
+SYNTHESIZER_USER = """USER QUESTION:
+{question}
+
+EXECUTION SUMMARY (Results from specialized workers):
+{worker_results}
+
+AVAILABLE ARTIFACTS:
+{artifacts_summary}
+
+Provide the final integrated response:"""
+
+SYNTHESIZER_PROMPT = ChatPromptTemplate.from_messages([
+    ("system", SYNTHESIZER_SYSTEM),
+    MessagesPlaceholder(variable_name="history"),
+    ("human", SYNTHESIZER_USER),
+])

backend/src/ea_chatbot/graph/state.py

@@ -12,12 +12,12 @@ class AgentState(AS):
     
     # Query Analysis (Decomposition results)
     analysis: Optional[Dict[str, Any]] 
-    # Expected keys: "requires_dataset", "expert", "data", "unknown", "condition"
+    # Expected keys: "data_required", "unknowns", "ambiguities", "conditions"
     
-    # Step-by-step reasoning
+    # Step-by-step reasoning (Legacy, use checklist for new flows)
     plan: Optional[str]
     
-    # Code execution context
+    # Code execution context (Legacy, use workers for new flows)
     code: Optional[str]
     code_output: Optional[str]
     error: Optional[str]
@@ -26,11 +26,21 @@ class AgentState(AS):
     plots: Annotated[List[Any], operator.add] # Matplotlib figures
     dfs: Dict[str, Any] # Pandas DataFrames
     
-    # Conversation summary
+    # Conversation summary / Latest worker result
     summary: Optional[str]
     
-    # Routing hint: "clarify", "plan", "research", "end"
+    # Routing hint: "clarify", "plan", "end"
     next_action: str
 
     # Number of execution attempts
     iterations: int
+
+    # --- DeepAgents Extensions ---
+    # High-level plan/checklist: List of Task objects (dicts)
+    checklist: List[Dict[str, Any]]
+    
+    # Current active step in the checklist
+    current_step: int
+    
+    # Virtual File System (VFS): Map of filenames to content/metadata
+    vfs: Dict[str, Any]

backend/src/ea_chatbot/graph/workers/data_analyst/mapping.py

@@ -0,0 +1,42 @@
+from typing import Dict, Any, List
+import copy
+from langchain_core.messages import HumanMessage, AIMessage
+from ea_chatbot.graph.state import AgentState
+from ea_chatbot.graph.workers.data_analyst.state import WorkerState
+from ea_chatbot.utils.vfs import safe_vfs_copy
+
+def prepare_worker_input(state: AgentState) -> Dict[str, Any]:
+    """Prepare the initial state for the Data Analyst worker."""
+    checklist = state.get("checklist", [])
+    current_step = state.get("current_step", 0)
+    
+    # Get the current task description
+    task_desc = "Analyze data" # Default
+    if 0 <= current_step < len(checklist):
+        task_desc = checklist[current_step].get("task", task_desc)
+    
+    return {
+        "task": task_desc,
+        "messages": [HumanMessage(content=task_desc)], # Start worker loop with the task
+        "vfs_state": safe_vfs_copy(state.get("vfs", {})),
+        "iterations": 0,
+        "plots": [],
+        "code": None,
+        "output": None,
+        "error": None,
+        "result": None
+    }
+
+def merge_worker_output(worker_state: WorkerState) -> Dict[str, Any]:
+    """Map worker results back to the global AgentState."""
+    result = worker_state.get("result") or "No result produced by data analyst worker."
+    
+    # We bubble up the summary as an AI message and update the global summary field
+    updates = {
+        "messages": [AIMessage(content=result)],
+        "summary": result,
+        "vfs": worker_state.get("vfs_state", {}),
+        "plots": worker_state.get("plots", [])
+    }
+    
+    return updates

backend/src/ea_chatbot/graph/workers/data_analyst/nodes/coder.py

@@ -0,0 +1,64 @@
+from typing import Dict, Any, List, Optional
+from ea_chatbot.graph.workers.data_analyst.state import WorkerState
+from ea_chatbot.config import Settings
+from ea_chatbot.utils.llm_factory import get_llm_model
+from ea_chatbot.utils import database_inspection
+from ea_chatbot.utils.logging import get_logger, LangChainLoggingHandler
+from ea_chatbot.graph.prompts.coder import CODE_GENERATOR_PROMPT
+from ea_chatbot.schemas import CodeGenerationResponse
+
+def coder_node(state: WorkerState) -> dict:
+    """Generate Python code based on the sub-task assigned to the worker."""
+    task = state["task"]
+    output = state.get("output", "None")
+    error = state.get("error", "None")
+    vfs_state = state.get("vfs_state", {})
+    
+    settings = Settings()
+    logger = get_logger("data_analyst_worker:coder")
+    
+    logger.info(f"Generating Python code for task: {task[:50]}...")
+    
+    # We can use the configured 'coder_llm' for this node
+    llm = get_llm_model(
+        settings.coder_llm,
+        callbacks=[LangChainLoggingHandler(logger=logger)]
+    )
+    structured_llm = llm.with_structured_output(CodeGenerationResponse)
+    
+    # Data summary for context
+    database_description = database_inspection.get_data_summary(data_dir=settings.data_dir) or "No data available."
+    
+    # VFS Summary: Let the LLM know which files are available in-memory
+    vfs_summary = "Available in-memory files (VFS):\n"
+    if vfs_state:
+        for filename, data in vfs_state.items():
+            if isinstance(data, dict):
+                meta = data.get("metadata", {})
+                vfs_summary += f"- {filename} ({meta.get('type', 'unknown')})\n"
+            else:
+                vfs_summary += f"- {filename} (raw data)\n"
+    else:
+        vfs_summary += "- None"
+
+    # Reuse the global prompt but adapt 'question' and 'plan' labels
+    # For a sub-task worker, 'task' effectively replaces the high-level 'plan'
+    messages = CODE_GENERATOR_PROMPT.format_messages(
+        question=task,
+        plan="Focus on the specific task below.",
+        database_description=database_description,
+        code_exec_results=f"Output: {output}\nError: {error}\n\n{vfs_summary}",
+        example_code=""
+    )
+    
+    try:
+        response = structured_llm.invoke(messages)
+        logger.info("[bold green]Code generated.[/bold green]")
+        return {
+            "code": response.parsed_code,
+            "error": None, # Clear previous errors
+            "iterations": state.get("iterations", 0) + 1
+        }
+    except Exception as e:
+        logger.error(f"Failed to generate code: {str(e)}")
+        raise e

backend/src/ea_chatbot/graph/workers/data_analyst/nodes/executor.py

@@ -1,23 +1,25 @@
 import io
 import sys
 import traceback
+import copy
 from contextlib import redirect_stdout
 from typing import TYPE_CHECKING
 import pandas as pd
 from matplotlib.figure import Figure
 
-from ea_chatbot.graph.state import AgentState
+from ea_chatbot.graph.workers.data_analyst.state import WorkerState
 from ea_chatbot.utils.db_client import DBClient
+from ea_chatbot.utils.vfs import VFSHelper, safe_vfs_copy
 from ea_chatbot.utils.logging import get_logger
 from ea_chatbot.config import Settings
 
 if TYPE_CHECKING:
     from ea_chatbot.types import DBSettings
 
-def executor_node(state: AgentState) -> dict:
-    """Execute the Python code and capture output, plots, and dataframes."""
+def executor_node(state: WorkerState) -> dict:
+    """Execute the Python code in the context of the Data Analyst worker."""
     code = state.get("code")
-    logger = get_logger("executor")
+    logger = get_logger("data_analyst_worker:executor")
     
     if not code:
         logger.error("No code provided to executor.")
@@ -37,45 +39,40 @@ def executor_node(state: AgentState) -> dict:
     
     db_client = DBClient(settings=db_settings)
     
+    # Initialize the Virtual File System (VFS) helper with the snapshot from state
+    vfs_state = safe_vfs_copy(state.get("vfs_state", {}))
+    vfs_helper = VFSHelper(vfs_state)
+    
     # Initialize local variables for execution
-    # 'db' is the DBClient instance, 'plots' is for matplotlib figures
     local_vars = {
         'db': db_client,
         'plots': [],
-        'pd': pd
+        'pd': pd,
+        'vfs': vfs_helper
     }
     
     stdout_buffer = io.StringIO()
     error = None
-    code_output = ""
+    output = ""
     plots = []
-    dfs = {}
 
     try:
         with redirect_stdout(stdout_buffer):
             # Execute the code in the context of local_vars
             exec(code, {}, local_vars)
             
-        code_output = stdout_buffer.getvalue()
+        output = stdout_buffer.getvalue()
         
         # Limit the output length if it's too long
-        if code_output.count('\n') > 32:
-            code_output = '\n'.join(code_output.split('\n')[:32]) + '\n...'
+        if output.count('\n') > 32:
+            output = '\n'.join(output.split('\n')[:32]) + '\n...'
 
         # Extract plots
         raw_plots = local_vars.get('plots', [])
         if isinstance(raw_plots, list):
             plots = [p for p in raw_plots if isinstance(p, Figure)]
 
-        # Extract DataFrames that were likely intended for display
-        # We look for DataFrames in local_vars that were mentioned in the code
-        for key, value in local_vars.items():
-            if isinstance(value, pd.DataFrame):
-                # Heuristic: if the variable name is in the code, it might be a result DF
-                if key in code:
-                    dfs[key] = value
-        
-        logger.info(f"[bold green]Execution complete.[/bold green] Captured {len(plots)} plots and {len(dfs)} dataframes.")
+        logger.info(f"[bold green]Execution complete.[/bold green] Captured {len(plots)} plots.")
 
     except Exception as e:
         # Capture the traceback
@@ -90,13 +87,11 @@ def executor_node(state: AgentState) -> dict:
         error += f"{exc_type.__name__ if exc_type else 'Exception'}: {exc_value}"
         
         logger.error(f"Execution failed: {str(e)}")
-        
-        # If we have an error, we still might want to see partial stdout
-        code_output = stdout_buffer.getvalue()
+        output = stdout_buffer.getvalue()
 
     return {
-        "code_output": code_output,
+        "output": output,
         "error": error,
         "plots": plots,
-        "dfs": dfs
+        "vfs_state": vfs_state
     }

backend/src/ea_chatbot/graph/workers/data_analyst/nodes/summarizer.py

@@ -0,0 +1,53 @@
+from typing import Dict, Any, List, Optional
+from ea_chatbot.graph.workers.data_analyst.state import WorkerState
+from ea_chatbot.config import Settings
+from ea_chatbot.utils.llm_factory import get_llm_model
+from ea_chatbot.utils.logging import get_logger, LangChainLoggingHandler
+
+def summarizer_node(state: WorkerState) -> dict:
+    """Summarize the data analysis results for the Orchestrator."""
+    task = state["task"]
+    output = state.get("output", "")
+    error = state.get("error")
+    plots = state.get("plots", [])
+    vfs_state = state.get("vfs_state", {})
+    
+    settings = Settings()
+    logger = get_logger("data_analyst_worker:summarizer")
+    
+    logger.info("Summarizing analysis results for the Orchestrator...")
+    
+    # Artifact summary
+    artifact_info = ""
+    if plots:
+        artifact_info += f"- Generated {len(plots)} plot(s).\n"
+    if vfs_state:
+        artifact_info += "- VFS Artifacts: " + ", ".join(vfs_state.keys()) + "\n"
+
+    # We can use a smaller/faster model for this summary if needed
+    llm = get_llm_model(
+        settings.planner_llm, # Using planner model for summary logic
+        callbacks=[LangChainLoggingHandler(logger=logger)]
+    )
+    
+    prompt = f"""You are a data analyst sub-agent. You have completed a sub-task.
+Task: {task}
+Execution Results: {output}
+Error Log (if any): {error}
+{artifact_info}
+
+Provide a concise summary of the findings or status for the top-level Orchestrator. 
+If plots or data files were generated, mention them.
+If the execution failed after multiple retries, explain why concisely. 
+Do NOT include the raw Python code, just the results of the analysis."""
+
+    try:
+        response = llm.invoke(prompt)
+        result = response.content if hasattr(response, "content") else str(response)
+        logger.info("[bold green]Analysis results summarized.[/bold green]")
+        return {
+            "result": result
+        }
+    except Exception as e:
+        logger.error(f"Failed to summarize results: {str(e)}")
+        raise e

backend/src/ea_chatbot/graph/workers/data_analyst/state.py

@@ -0,0 +1,28 @@
+from typing import TypedDict, List, Dict, Any, Optional
+from langchain_core.messages import BaseMessage
+
+class WorkerState(TypedDict):
+    """Internal state for the Data Analyst worker subgraph."""
+    
+    # Internal worker conversation (not bubbled up to global unless summary)
+    messages: List[BaseMessage]
+    
+    # The specific sub-task assigned by the Orchestrator
+    task: str
+    
+    # Generated code and execution context
+    code: Optional[str]
+    output: Optional[str]
+    error: Optional[str]
+    
+    # Number of internal retry attempts
+    iterations: int
+    
+    # Temporary storage for analysis results
+    plots: List[Any]
+    
+    # Isolated/Snapshot view of the VFS
+    vfs_state: Dict[str, Any]
+    
+    # The final summary or result to return to the Orchestrator
+    result: Optional[str]

backend/src/ea_chatbot/graph/workers/data_analyst/workflow.py

@@ -0,0 +1,51 @@
+from langgraph.graph import StateGraph, END
+from langgraph.graph.state import CompiledStateGraph
+from ea_chatbot.graph.workers.data_analyst.state import WorkerState
+from ea_chatbot.graph.workers.data_analyst.nodes.coder import coder_node
+from ea_chatbot.graph.workers.data_analyst.nodes.executor import executor_node
+from ea_chatbot.graph.workers.data_analyst.nodes.summarizer import summarizer_node
+
+def router(state: WorkerState) -> str:
+    """Routes the subgraph between coding, execution, and summarization."""
+    error = state.get("error")
+    iterations = state.get("iterations", 0)
+    
+    if error and iterations < 3:
+        # Retry with error correction
+        return "coder"
+    
+    # Either success or max retries reached
+    return "summarizer"
+
+def create_data_analyst_worker(
+    coder=coder_node,
+    executor=executor_node,
+    summarizer=summarizer_node
+) -> CompiledStateGraph:
+    """Create the Data Analyst worker subgraph."""
+    workflow = StateGraph(WorkerState)
+    
+    # Add Nodes
+    workflow.add_node("coder", coder)
+    workflow.add_node("executor", executor)
+    workflow.add_node("summarizer", summarizer)
+    
+    # Set entry point
+    workflow.set_entry_point("coder")
+    
+    # Add Edges
+    workflow.add_edge("coder", "executor")
+    
+    # Add Conditional Edges
+    workflow.add_conditional_edges(
+        "executor",
+        router,
+        {
+            "coder": "coder",
+            "summarizer": "summarizer"
+        }
+    )
+    
+    workflow.add_edge("summarizer", END)
+    
+    return workflow.compile()

backend/src/ea_chatbot/graph/workers/researcher/mapping.py

@@ -0,0 +1,32 @@
+from typing import Dict, Any, List
+from langchain_core.messages import HumanMessage, AIMessage
+from ea_chatbot.graph.state import AgentState
+from ea_chatbot.graph.workers.researcher.state import WorkerState
+
+def prepare_researcher_input(state: AgentState) -> Dict[str, Any]:
+    """Prepare the initial state for the Researcher worker."""
+    checklist = state.get("checklist", [])
+    current_step = state.get("current_step", 0)
+    
+    task_desc = "Perform research"
+    if 0 <= current_step < len(checklist):
+        task_desc = checklist[current_step].get("task", task_desc)
+    
+    return {
+        "task": task_desc,
+        "messages": [HumanMessage(content=task_desc)],
+        "queries": [],
+        "raw_results": [],
+        "iterations": 0,
+        "result": None
+    }
+
+def merge_researcher_output(worker_state: WorkerState) -> Dict[str, Any]:
+    """Map researcher results back to the global AgentState."""
+    result = worker_state.get("result") or "No result produced by researcher worker."
+    
+    return {
+        "messages": [AIMessage(content=result)],
+        "summary": result,
+        # Researcher doesn't usually update VFS or Plots, but we keep the structure
+    }

backend/src/ea_chatbot/graph/workers/researcher/nodes/searcher.py

@@ -1,24 +1,20 @@
 from langchain_openai import ChatOpenAI
 from langchain_google_genai import ChatGoogleGenerativeAI
-from ea_chatbot.graph.state import AgentState
+from ea_chatbot.graph.workers.researcher.state import WorkerState
 from ea_chatbot.config import Settings
 from ea_chatbot.utils.llm_factory import get_llm_model
 from ea_chatbot.utils import helpers
 from ea_chatbot.utils.logging import get_logger, LangChainLoggingHandler
 from ea_chatbot.graph.prompts.researcher import RESEARCHER_PROMPT
 
-def researcher_node(state: AgentState) -> dict:
-    """Handle general research queries or web searches."""
-    question = state["question"]
-    history = state.get("messages", [])[-6:]
-    summary = state.get("summary", "")
+def searcher_node(state: WorkerState) -> dict:
+    """Execute web research for the specific task."""
+    task = state["task"]
+    logger = get_logger("researcher_worker:searcher")
+    
+    logger.info(f"Researching task: {task[:50]}...")
     
     settings = Settings()
-    logger = get_logger("researcher")
-    
-    logger.info(f"Researching question: [italic]\"{question}\"[/italic]")
-    
-    # Use researcher_llm from settings
     llm = get_llm_model(
         settings.researcher_llm,
         callbacks=[LangChainLoggingHandler(logger=logger)]
@@ -26,34 +22,39 @@ def researcher_node(state: AgentState) -> dict:
     
     date_str = helpers.get_readable_date()
     
+    # Adapt the global researcher prompt for the sub-task
     messages = RESEARCHER_PROMPT.format_messages(
         date=date_str,
-        question=question,
-        history=history,
-        summary=summary
+        question=task,
+        history=[], # Worker has fresh context or task-specific history
+        summary=""
     )
     
-    # Provider-aware tool binding
+    # Tool binding
     try:
         if isinstance(llm, ChatGoogleGenerativeAI):
-            # Native Google Search for Gemini
             llm_with_tools = llm.bind_tools([{"google_search": {}}])
         elif isinstance(llm, ChatOpenAI):
-            # Native Web Search for OpenAI (built-in tool)
             llm_with_tools = llm.bind_tools([{"type": "web_search"}])
         else:
-            # Fallback for other providers that might not support these specific search tools
             llm_with_tools = llm
     except Exception as e:
-        logger.warning(f"Failed to bind search tools: {str(e)}. Falling back to base LLM.")
+        logger.warning(f"Failed to bind search tools: {str(e)}")
         llm_with_tools = llm
     
     try:
         response = llm_with_tools.invoke(messages)
-        logger.info("[bold green]Research complete.[/bold green]")
+        logger.info("[bold green]Search complete.[/bold green]")
+        
+        # In a real tool-use scenario, we'd extract the tool outputs here.
+        # For now, we'll store the response content as a 'raw_result'.
+        content = response.content if hasattr(response, "content") else str(response)
+        
         return {
-            "messages": [response]
+            "messages": [response],
+            "raw_results": [content],
+            "iterations": state.get("iterations", 0) + 1
         }
     except Exception as e:
-        logger.error(f"Research failed: {str(e)}")
-        raise e
+        logger.error(f"Search failed: {str(e)}")
+        raise e

backend/src/ea_chatbot/graph/workers/researcher/nodes/summarizer.py

@@ -0,0 +1,49 @@
+from ea_chatbot.graph.workers.researcher.state import WorkerState
+from ea_chatbot.config import Settings
+from ea_chatbot.utils.llm_factory import get_llm_model
+from ea_chatbot.utils.logging import get_logger, LangChainLoggingHandler
+
+def summarizer_node(state: WorkerState) -> dict:
+    """Summarize research results for the Orchestrator."""
+    task = state["task"]
+    raw_results = state.get("raw_results", [])
+    
+    logger = get_logger("researcher_worker:summarizer")
+    logger.info("Summarizing research results...")
+    
+    settings = Settings()
+    llm = get_llm_model(
+        settings.planner_llm,
+        callbacks=[LangChainLoggingHandler(logger=logger)]
+    )
+    
+    # Ensure all results are strings (Gemini/OpenAI might return complex content)
+    processed_results = []
+    for res in raw_results:
+        if isinstance(res, list):
+            processed_results.append(str(res))
+        else:
+            processed_results.append(str(res))
+
+    results_str = "\n---\n".join(processed_results)
+    
+    prompt = f"""You are a Research Specialist sub-agent. You have completed a research sub-task.
+Task: {task}
+
+Raw Research Findings:
+{results_str}
+
+Provide a concise, factual summary of the findings for the top-level Orchestrator.
+Ensure all key facts, dates, and sources (if provided) are preserved.
+Do NOT include internal reasoning, just the factual summary."""
+
+    try:
+        response = llm.invoke(prompt)
+        result = response.content if hasattr(response, "content") else str(response)
+        logger.info("[bold green]Research summary complete.[/bold green]")
+        return {
+            "result": result
+        }
+    except Exception as e:
+        logger.error(f"Failed to summarize research: {str(e)}")
+        raise e

backend/src/ea_chatbot/graph/workers/researcher/state.py

@@ -0,0 +1,21 @@
+from typing import TypedDict, List, Dict, Any, Optional
+from langchain_core.messages import BaseMessage
+
+class WorkerState(TypedDict):
+    """Internal state for the Researcher worker subgraph."""
+    
+    # Internal worker conversation
+    messages: List[BaseMessage]
+    
+    # The specific sub-task assigned by the Orchestrator
+    task: str
+    
+    # Search context
+    queries: List[str]
+    raw_results: List[str]
+    
+    # Number of internal retry/refinement attempts
+    iterations: int
+    
+    # The final research summary to return to the Orchestrator
+    result: Optional[str]

backend/src/ea_chatbot/graph/workers/researcher/workflow.py

@@ -0,0 +1,25 @@
+from langgraph.graph import StateGraph, END
+from langgraph.graph.state import CompiledStateGraph
+from ea_chatbot.graph.workers.researcher.state import WorkerState
+from ea_chatbot.graph.workers.researcher.nodes.searcher import searcher_node
+from ea_chatbot.graph.workers.researcher.nodes.summarizer import summarizer_node
+
+def create_researcher_worker(
+    searcher=searcher_node,
+    summarizer=summarizer_node
+) -> CompiledStateGraph:
+    """Create the Researcher worker subgraph."""
+    workflow = StateGraph(WorkerState)
+    
+    # Add Nodes
+    workflow.add_node("searcher", searcher)
+    workflow.add_node("summarizer", summarizer)
+    
+    # Set entry point
+    workflow.set_entry_point("searcher")
+    
+    # Add Edges
+    workflow.add_edge("searcher", "summarizer")
+    workflow.add_edge("summarizer", END)
+    
+    return workflow.compile()

backend/src/ea_chatbot/graph/workflow.py

@@ -2,86 +2,101 @@ from langgraph.graph import StateGraph, END
 from ea_chatbot.graph.state import AgentState
 from ea_chatbot.graph.nodes.query_analyzer import query_analyzer_node
 from ea_chatbot.graph.nodes.planner import planner_node
-from ea_chatbot.graph.nodes.coder import coder_node
-from ea_chatbot.graph.nodes.error_corrector import error_corrector_node
-from ea_chatbot.graph.nodes.executor import executor_node
-from ea_chatbot.graph.nodes.summarizer import summarizer_node
-from ea_chatbot.graph.nodes.researcher import researcher_node
+from ea_chatbot.graph.nodes.delegate import delegate_node
+from ea_chatbot.graph.nodes.reflector import reflector_node
+from ea_chatbot.graph.nodes.synthesizer import synthesizer_node
+from ea_chatbot.graph.workers.data_analyst.workflow import create_data_analyst_worker
+from ea_chatbot.graph.workers.data_analyst.mapping import prepare_worker_input, merge_worker_output
+from ea_chatbot.graph.workers.researcher.workflow import create_researcher_worker
+from ea_chatbot.graph.workers.researcher.mapping import prepare_researcher_input, merge_researcher_output
 from ea_chatbot.graph.nodes.clarification import clarification_node
 from ea_chatbot.graph.nodes.summarize_conversation import summarize_conversation_node
 
-MAX_ITERATIONS = 3
+# Global cache for compiled subgraphs
+_DATA_ANALYST_WORKER = create_data_analyst_worker()
+_RESEARCHER_WORKER = create_researcher_worker()
 
-def router(state: AgentState) -> str:
-    """Route to the next node based on the analysis."""
+# Define worker nodes as piped runnables to enable subgraph event propagation
+data_analyst_worker_runnable = prepare_worker_input | _DATA_ANALYST_WORKER | merge_worker_output
+researcher_worker_runnable = prepare_researcher_input | _RESEARCHER_WORKER | merge_researcher_output
+
+def main_router(state: AgentState) -> str:
+    """Route from query analyzer based on initial assessment."""
     next_action = state.get("next_action")
-    if next_action == "plan":
-        return "planner"
-    elif next_action == "research":
-        return "researcher"
-    elif next_action == "clarify":
+    if next_action == "clarify":
         return "clarification"
-    else:
-        return END
+    # Even if QA suggests 'research', we now go through 'planner' for orchestration
+    # aligning with the new hierarchical architecture.
+    return "planner"
 
-def create_workflow():
-    """Create the LangGraph workflow."""
+def delegation_router(state: AgentState) -> str:
+    """Route from delegate node to specific workers or synthesis."""
+    next_action = state.get("next_action")
+    if next_action == "data_analyst":
+        return "data_analyst_worker"
+    elif next_action == "researcher":
+        return "researcher_worker"
+    elif next_action == "summarize":
+        return "synthesizer"
+    return "synthesizer"
+
+def create_workflow(
+    query_analyzer=query_analyzer_node,
+    planner=planner_node,
+    delegate=delegate_node,
+    data_analyst_worker=data_analyst_worker_runnable,
+    researcher_worker=researcher_worker_runnable,
+    reflector=reflector_node,
+    synthesizer=synthesizer_node,
+    clarification=clarification_node,
+    summarize_conversation=summarize_conversation_node
+):
+    """Create the high-level Orchestrator workflow."""
     workflow = StateGraph(AgentState)
     
-    # Add nodes
-    workflow.add_node("query_analyzer", query_analyzer_node)
-    workflow.add_node("planner", planner_node)
-    workflow.add_node("coder", coder_node)
-    workflow.add_node("error_corrector", error_corrector_node)
-    workflow.add_node("researcher", researcher_node)
-    workflow.add_node("clarification", clarification_node)
-    workflow.add_node("executor", executor_node)
-    workflow.add_node("summarizer", summarizer_node)
-    workflow.add_node("summarize_conversation", summarize_conversation_node)
+    # Add Nodes
+    workflow.add_node("query_analyzer", query_analyzer)
+    workflow.add_node("planner", planner)
+    workflow.add_node("delegate", delegate)
+    workflow.add_node("data_analyst_worker", data_analyst_worker)
+    workflow.add_node("researcher_worker", researcher_worker)
+    workflow.add_node("reflector", reflector)
+    workflow.add_node("synthesizer", synthesizer)
+    workflow.add_node("clarification", clarification)
+    workflow.add_node("summarize_conversation", summarize_conversation)
     
     # Set entry point
     workflow.set_entry_point("query_analyzer")
     
-    # Add conditional edges from query_analyzer
+    # Edges
     workflow.add_conditional_edges(
         "query_analyzer",
-        router,
+        main_router,
         {
-            "planner": "planner",
-            "researcher": "researcher",
             "clarification": "clarification",
-            END: END
+            "planner": "planner"
         }
     )
     
-    # Linear flow for planning and coding
-    workflow.add_edge("planner", "coder")
-    workflow.add_edge("coder", "executor")
+    workflow.add_edge("planner", "delegate")
     
-    # Executor routing
-    def executor_router(state: AgentState) -> str:
-        if state.get("error"):
-            # Check for iteration limit to prevent infinite loops
-            if state.get("iterations", 0) >= MAX_ITERATIONS:
-                return "summarizer"
-            return "error_corrector"
-        return "summarizer"
-        
     workflow.add_conditional_edges(
-        "executor",
-        executor_router,
+        "delegate",
+        delegation_router,
         {
-            "error_corrector": "error_corrector",
-            "summarizer": "summarizer"
+            "data_analyst_worker": "data_analyst_worker",
+            "researcher_worker": "researcher_worker",
+            "synthesizer": "synthesizer"
         }
     )
     
-    workflow.add_edge("error_corrector", "executor")
+    workflow.add_edge("data_analyst_worker", "reflector")
+    workflow.add_edge("researcher_worker", "reflector")
+    workflow.add_edge("reflector", "delegate")
     
-    workflow.add_edge("researcher", "summarize_conversation")
-    workflow.add_edge("clarification", END)
-    workflow.add_edge("summarizer", "summarize_conversation")
+    workflow.add_edge("synthesizer", "summarize_conversation")
     workflow.add_edge("summarize_conversation", END)
+    workflow.add_edge("clarification", END)
     
     # Compile the graph
     app = workflow.compile()

backend/src/ea_chatbot/schemas.py

@@ -1,7 +1,15 @@
 from pydantic import BaseModel, Field, computed_field
-from typing import Sequence, Optional
+from typing import Sequence, Optional, List, Dict, Any, Literal
 import re
 
+class QueryAnalysis(BaseModel):
+    """Analysis of the user's query."""
+    data_required: List[str] = Field(description="List of data points or entities mentioned (e.g., ['2024 results', 'Florida']).")
+    unknowns: List[str] = Field(description="List of target information the user wants to know or needed for final answer (e.g., 'who won', 'total votes').")
+    ambiguities: List[str] = Field(description="List of CRITICAL missing details that prevent ANY analysis. Do NOT include database names or plot types if defaults can be used.")
+    conditions: List[str] = Field(description="List of any filters or constraints (e.g., ['year=2024', 'state=Florida']). Include context resolved from history.")
+    next_action: Literal["plan", "clarify", "research"] = Field(description="The next action to take. 'plan' for data analysis (even with defaults), 'research' for general knowledge, or 'clarify' ONLY for critical ambiguities.")
+
 class TaskPlanContext(BaseModel):
     '''Background context relevant to the task plan'''
     initial_context: str = Field(
@@ -33,6 +41,22 @@ class TaskPlanResponse(BaseModel):
         description="Ordered list of steps to execute that follow the 'Step <number>: <detail>' pattern.",
     )
 
+class ChecklistTask(BaseModel):
+    '''A specific sub-task in the high-level orchestrator plan'''
+    task: str = Field(description="Description of the sub-task")
+    worker: str = Field(description="The worker to delegate to (data_analyst or researcher)")
+
+class ChecklistResponse(BaseModel):
+    '''Orchestrator's decomposed plan/checklist'''
+    goal: str = Field(description="Overall objective")
+    reflection: str = Field(description="Strategic reasoning")
+    checklist: List[ChecklistTask] = Field(description="Ordered list of tasks for specialized workers")
+
+class ReflectorResponse(BaseModel):
+    '''Orchestrator's evaluation of worker output'''
+    satisfied: bool = Field(description="Whether the worker's output satisfies the sub-task requirements")
+    reasoning: str = Field(description="Brief explanation of the evaluation")
+
 _IM_SEP_TOKEN_PATTERN = re.compile(re.escape("<|im_sep|>"))
 _CODE_BLOCK_PATTERN = re.compile(r"```(?:python\s*)?(.*?)\s*```", re.DOTALL)
 _FORBIDDEN_MODULES = (

backend/src/ea_chatbot/utils/vfs.py

@@ -0,0 +1,69 @@
+import copy
+from typing import Dict, Any, Optional, Tuple, List
+from ea_chatbot.utils.logging import get_logger
+
+logger = get_logger("utils:vfs")
+
+def safe_vfs_copy(vfs_state: Dict[str, Any]) -> Dict[str, Any]:
+    """
+    Perform a safe deep copy of the VFS state.
+    
+    If an entry cannot be deep-copied (e.g., it contains a non-copyable object like a DB handle),
+    logs an error and replaces the entry with a descriptive error marker.
+    This prevents crashing the graph/persistence while making the failure explicit.
+    """
+    new_vfs = {}
+    for filename, data in vfs_state.items():
+        try:
+            # Attempt a standard deepcopy for isolation
+            new_vfs[filename] = copy.deepcopy(data)
+        except Exception as e:
+            logger.error(
+                f"CRITICAL: VFS artifact '{filename}' is NOT copyable/serializable: {str(e)}. "
+                "Replacing with error placeholder to prevent graph crash."
+            )
+            # Replace with a standardized error artifact
+            new_vfs[filename] = {
+                "content": f"<ERROR: This artifact could not be persisted or copied: {str(e)}>",
+                "metadata": {
+                    "type": "error",
+                    "error": str(e),
+                    "original_filename": filename
+                }
+            }
+    return new_vfs
+
+class VFSHelper:
+    """Helper class for managing in-memory Virtual File System (VFS) artifacts."""
+    
+    def __init__(self, vfs_state: Dict[str, Any]):
+        """Initialize with a reference to the VFS state from AgentState."""
+        self._vfs = vfs_state
+        
+    def write(self, filename: str, content: Any, metadata: Optional[Dict[str, Any]] = None) -> None:
+        """Write a file to the VFS."""
+        self._vfs[filename] = {
+            "content": content,
+            "metadata": metadata or {}
+        }
+        
+    def read(self, filename: str) -> Tuple[Optional[Any], Optional[Dict[str, Any]]]:
+        """Read a file and its metadata from the VFS. Returns (None, None) if not found."""
+        file_data = self._vfs.get(filename)
+        if file_data is not None:
+            # Handle raw values (backwards compatibility or inconsistent schema)
+            if not isinstance(file_data, dict) or "content" not in file_data:
+                return file_data, {}
+            return file_data["content"], file_data.get("metadata", {})
+        return None, None
+        
+    def list(self) -> List[str]:
+        """List all filenames in the VFS."""
+        return list(self._vfs.keys())
+        
+    def delete(self, filename: str) -> bool:
+        """Delete a file from the VFS. Returns True if deleted, False if not found."""
+        if filename in self._vfs:
+            del self._vfs[filename]
+            return True
+        return False

backend/tests/api/test_agent_stream.py

@@ -36,24 +36,25 @@ async def test_stream_agent_events_all_features():
         # Stream chunk
         {
             "event": "on_chat_model_stream", 
-            "metadata": {"langgraph_node": "summarizer"}, 
+            "metadata": {"langgraph_node": "synthesizer"}, 
             "data": {"chunk": AIMessage(content="Hello ")}
         },
         {
             "event": "on_chat_model_stream", 
-            "metadata": {"langgraph_node": "summarizer"}, 
+            "metadata": {"langgraph_node": "synthesizer"}, 
             "data": {"chunk": AIMessage(content="world")}
         },
-        # Plot event
+        # Plot event - with nested subgraph it might bubble up or come directly from data_analyst_worker
+        # Let's mock it coming from the data_analyst_worker on_chain_end
         {
             "event": "on_chain_end", 
-            "name": "executor", 
+            "name": "data_analyst_worker", 
             "data": {"output": {"plots": [fig]}}
         },
         # Final response
         {
             "event": "on_chain_end", 
-            "name": "summarizer", 
+            "name": "synthesizer", 
             "data": {"output": {"messages": [AIMessage(content="Hello world final")]}}
         },
         # Summary update
@@ -91,7 +92,7 @@ async def test_stream_agent_events_all_features():
         assert any(r.get("type") == "on_chat_model_stream" for r in results)
         
         # Verify plot was encoded
-        plot_event = next(r for r in results if r.get("name") == "executor")
+        plot_event = next(r for r in results if r.get("name") == "data_analyst_worker")
         assert "encoded_plots" in plot_event["data"]
         assert len(plot_event["data"]["encoded_plots"]) == 1
         

backend/tests/api/test_api_auth.py

@@ -123,3 +123,20 @@ def test_get_me_success(client):
         assert response.status_code == 200
         assert response.json()["email"] == "test@example.com"
         assert response.json()["id"] == "123"
+
+def test_get_me_rejects_refresh_token(client):
+    """Test that /auth/me rejects refresh tokens for authentication."""
+    from ea_chatbot.api.utils import create_refresh_token
+    token = create_refresh_token(data={"sub": "123"})
+    
+    with patch("ea_chatbot.api.dependencies.history_manager") as mock_hm:
+        # Even if the user exists, the dependency should reject the token type
+        mock_hm.get_user_by_id.return_value = User(id="123", username="test@example.com")
+        
+        response = client.get(
+            "/api/v1/auth/me",
+            headers={"Authorization": f"Bearer {token}"}
+        )
+        
+        assert response.status_code == 401
+        assert "Cannot use refresh token" in response.json()["detail"]

backend/tests/api/test_persistence.py

@@ -19,14 +19,13 @@ def auth_header(mock_user):
     yield {"Authorization": f"Bearer {token}"}
     app.dependency_overrides.clear()
 
-def test_persistence_integration_success(auth_header, mock_user):
-    """Test that messages and plots are persisted correctly during streaming."""
-    mock_events = [
-        {"event": "on_chat_model_stream", "name": "summarizer", "data": {"chunk": "Final answer"}},
-        {"event": "on_chain_end", "name": "summarizer", "data": {"output": {"messages": [{"content": "Final answer"}]}}},
-        {"event": "on_chain_end", "name": "summarize_conversation", "data": {"output": {"summary": "New summary"}}}
-    ]
-    
+    def test_persistence_integration_success(auth_header, mock_user):
+        """Test that messages and plots are persisted correctly during streaming."""
+        mock_events = [
+            {"event": "on_chat_model_stream", "metadata": {"langgraph_node": "synthesizer"}, "data": {"chunk": "Final answer"}},
+            {"event": "on_chain_end", "name": "synthesizer", "data": {"output": {"messages": [{"content": "Final answer"}]}}},
+            {"event": "on_chain_end", "name": "summarize_conversation", "data": {"output": {"summary": "New summary"}}}
+        ]
     async def mock_astream_events(*args, **kwargs):
         for event in mock_events:
             yield event

backend/tests/test_coder.py

@@ -1,62 +0,0 @@
-import pytest
-from unittest.mock import MagicMock, patch
-from ea_chatbot.graph.nodes.coder import coder_node
-from ea_chatbot.graph.nodes.error_corrector import error_corrector_node
-
-@pytest.fixture
-def mock_state():
-    return {
-        "messages": [],
-        "question": "Show me results for New Jersey",
-        "plan": "Step 1: Load data\nStep 2: Filter by NJ",
-        "code": None,
-        "error": None,
-        "plots": [],
-        "dfs": {},
-        "next_action": "plan"
-    }
-
-@patch("ea_chatbot.graph.nodes.coder.get_llm_model")
-@patch("ea_chatbot.utils.database_inspection.get_data_summary")
-def test_coder_node(mock_get_summary, mock_get_llm, mock_state):
-    """Test coder node generates code from plan."""
-    mock_get_summary.return_value = "Column: Name, Type: text"
-    
-    mock_llm = MagicMock()
-    mock_get_llm.return_value = mock_llm
-    
-    from ea_chatbot.schemas import CodeGenerationResponse
-    mock_response = CodeGenerationResponse(
-        code="import pandas as pd\nprint('Hello')",
-        explanation="Generated code"
-    )
-    mock_llm.with_structured_output.return_value.invoke.return_value = mock_response
-    
-    result = coder_node(mock_state)
-    
-    assert "code" in result
-    assert "import pandas as pd" in result["code"]
-    assert "error" in result
-    assert result["error"] is None
-
-@patch("ea_chatbot.graph.nodes.error_corrector.get_llm_model")
-def test_error_corrector_node(mock_get_llm, mock_state):
-    """Test error corrector node fixes code."""
-    mock_state["code"] = "import pandas as pd\nprint(undefined_var)"
-    mock_state["error"] = "NameError: name 'undefined_var' is not defined"
-    
-    mock_llm = MagicMock()
-    mock_get_llm.return_value = mock_llm
-    
-    from ea_chatbot.schemas import CodeGenerationResponse
-    mock_response = CodeGenerationResponse(
-        code="import pandas as pd\nprint('Defined')",
-        explanation="Fixed variable"
-    )
-    mock_llm.with_structured_output.return_value.invoke.return_value = mock_response
-    
-    result = error_corrector_node(mock_state)
-    
-    assert "code" in result
-    assert "print('Defined')" in result["code"]
-    assert result["error"] is None

backend/tests/test_data_analyst_mapping.py

@@ -0,0 +1,53 @@
+from langchain_core.messages import HumanMessage, AIMessage
+from ea_chatbot.graph.state import AgentState
+from ea_chatbot.graph.workers.data_analyst.mapping import (
+    prepare_worker_input,
+    merge_worker_output
+)
+from ea_chatbot.graph.workers.data_analyst.state import WorkerState
+
+def test_prepare_worker_input():
+    """Verify that we correctly map global state to worker input."""
+    global_state = AgentState(
+        messages=[HumanMessage(content="global message")],
+        question="original question",
+        checklist=[{"task": "Worker Task", "status": "pending"}],
+        current_step=0,
+        vfs={"old.txt": "old data"},
+        plots=[],
+        dfs={},
+        next_action="test",
+        iterations=0
+    )
+    
+    worker_input = prepare_worker_input(global_state)
+    
+    assert worker_input["task"] == "Worker Task"
+    assert "old.txt" in worker_input["vfs_state"]
+    # Internal worker messages should start fresh or with the task
+    assert len(worker_input["messages"]) == 1
+    assert worker_input["messages"][0].content == "Worker Task"
+
+def test_merge_worker_output():
+    """Verify that we correctly merge worker results back to global state."""
+    worker_state = WorkerState(
+        messages=[HumanMessage(content="internal"), AIMessage(content="summary")],
+        task="Worker Task",
+        result="Finished analysis",
+        plots=["plot1"],
+        vfs_state={"new.txt": "new data"},
+        iterations=2
+    )
+    
+    updates = merge_worker_output(worker_state)
+    
+    # We expect the 'result' to be added as an AI message to global history
+    assert len(updates["messages"]) == 1
+    assert updates["messages"][0].content == "Finished analysis"
+    
+    # VFS should be updated
+    assert "new.txt" in updates["vfs"]
+    
+    # Plots should be bubbled up
+    assert len(updates["plots"]) == 1
+    assert updates["plots"][0] == "plot1"

backend/tests/test_data_analyst_state.py

@@ -0,0 +1,17 @@
+from typing import get_type_hints, List
+from langchain_core.messages import BaseMessage
+from ea_chatbot.graph.workers.data_analyst.state import WorkerState
+
+def test_data_analyst_worker_state():
+    """Verify that DataAnalyst WorkerState has the required fields."""
+    hints = get_type_hints(WorkerState)
+    
+    assert "messages" in hints
+    assert "task" in hints
+    assert hints["task"] == str
+    
+    assert "code" in hints
+    assert "output" in hints
+    assert "error" in hints
+    assert "iterations" in hints
+    assert hints["iterations"] == int

backend/tests/test_data_analyst_worker.py

@@ -0,0 +1,81 @@
+import pytest
+from unittest.mock import MagicMock
+from ea_chatbot.graph.workers.data_analyst.workflow import create_data_analyst_worker, WorkerState
+
+def test_data_analyst_worker_one_shot():
+    """Verify a successful one-shot execution of the worker subgraph."""
+    mock_coder = MagicMock()
+    mock_executor = MagicMock()
+    mock_summarizer = MagicMock()
+    
+    # Scenario: Coder -> Executor (Success) -> Summarizer -> END
+    mock_coder.return_value = {"code": "print(1)", "error": None, "iterations": 1}
+    mock_executor.return_value = {"output": "1\n", "error": None, "plots": []}
+    mock_summarizer.return_value = {"result": "Result is 1"}
+    
+    graph = create_data_analyst_worker(
+        coder=mock_coder,
+        executor=mock_executor,
+        summarizer=mock_summarizer
+    )
+    
+    initial_state = WorkerState(
+        messages=[],
+        task="Calculate 1+1",
+        code=None,
+        output=None,
+        error=None,
+        iterations=0,
+        plots=[],
+        vfs_state={},
+        result=None
+    )
+    
+    final_state = graph.invoke(initial_state)
+    
+    assert final_state["result"] == "Result is 1"
+    assert mock_coder.call_count == 1
+    assert mock_executor.call_count == 1
+    assert mock_summarizer.call_count == 1
+
+def test_data_analyst_worker_retry():
+    """Verify that the worker retries on error."""
+    mock_coder = MagicMock()
+    mock_executor = MagicMock()
+    mock_summarizer = MagicMock()
+    
+    # Scenario: Coder (1) -> Executor (Error) -> Router (coder) -> Coder (2) -> Executor (Success) -> Summarizer -> END
+    mock_coder.side_effect = [
+        {"code": "error_code", "error": None, "iterations": 1},
+        {"code": "fixed_code", "error": None, "iterations": 2}
+    ]
+    mock_executor.side_effect = [
+        {"output": "", "error": "NameError", "plots": []},
+        {"output": "Success", "error": None, "plots": []}
+    ]
+    mock_summarizer.return_value = {"result": "Fixed Result"}
+    
+    graph = create_data_analyst_worker(
+        coder=mock_coder,
+        executor=mock_executor,
+        summarizer=mock_summarizer
+    )
+    
+    initial_state = WorkerState(
+        messages=[],
+        task="Retry Task",
+        code=None,
+        output=None,
+        error=None,
+        iterations=0,
+        plots=[],
+        vfs_state={},
+        result=None
+    )
+    
+    final_state = graph.invoke(initial_state)
+    
+    assert final_state["result"] == "Fixed Result"
+    assert mock_coder.call_count == 2
+    assert mock_executor.call_count == 2
+    assert mock_summarizer.call_count == 1

backend/tests/test_deepagents_e2e.py

@@ -0,0 +1,79 @@
+import pytest
+from unittest.mock import MagicMock
+from ea_chatbot.graph.workflow import create_workflow
+from ea_chatbot.graph.state import AgentState
+from langchain_core.messages import AIMessage, HumanMessage
+
+def test_deepagents_multi_worker_sequential_flow():
+    """Verify that the Orchestrator can handle a sequence of different workers."""
+    
+    mock_analyzer = MagicMock()
+    mock_planner = MagicMock()
+    mock_delegate = MagicMock()
+    mock_analyst = MagicMock()
+    mock_researcher = MagicMock()
+    mock_reflector = MagicMock()
+    mock_synthesizer = MagicMock()
+    
+    # 1. Analyzer: Plan
+    mock_analyzer.return_value = {"next_action": "plan"}
+    
+    # 2. Planner: Two tasks
+    mock_planner.return_value = {
+        "checklist": [
+            {"task": "Get Numbers", "worker": "data_analyst"},
+            {"task": "Get Facts", "worker": "researcher"}
+        ],
+        "current_step": 0
+    }
+    
+    # 3. Delegate & Reflector: Loop through tasks
+    mock_delegate.side_effect = [
+        {"next_action": "data_analyst"}, # Step 0
+        {"next_action": "researcher"},   # Step 1
+        {"next_action": "summarize"}     # Step 2 (done)
+    ]
+    
+    mock_analyst.return_value = {"messages": [AIMessage(content="Analyst summary")], "vfs": {"data.csv": "..."}}
+    mock_researcher.return_value = {"messages": [AIMessage(content="Researcher summary")]}
+    
+    mock_reflector.side_effect = [
+        {"current_step": 1, "next_action": "delegate"}, # Done with Analyst
+        {"current_step": 2, "next_action": "delegate"}  # Done with Researcher
+    ]
+    
+    mock_synthesizer.return_value = {
+        "messages": [AIMessage(content="Final multi-agent response")],
+        "next_action": "end"
+    }
+    
+    app = create_workflow(
+        query_analyzer=mock_analyzer,
+        planner=mock_planner,
+        delegate=mock_delegate,
+        data_analyst_worker=mock_analyst,
+        researcher_worker=mock_researcher,
+        reflector=mock_reflector,
+        synthesizer=mock_synthesizer
+    )
+    
+    initial_state = AgentState(
+        messages=[HumanMessage(content="Numbers and facts please")],
+        question="Numbers and facts please",
+        analysis={},
+        next_action="",
+        iterations=0,
+        checklist=[],
+        current_step=0,
+        vfs={},
+        plots=[],
+        dfs={}
+    )
+    
+    final_state = app.invoke(initial_state, config={"recursion_limit": 30})
+    
+    assert mock_analyst.called
+    assert mock_researcher.called
+    assert mock_reflector.call_count == 2
+    assert "Final multi-agent response" in [m.content for m in final_state["messages"]]
+    assert final_state["current_step"] == 2

backend/tests/test_executor.py

@@ -1,122 +0,0 @@
-import pytest
-import pandas as pd
-from unittest.mock import MagicMock, patch
-from matplotlib.figure import Figure
-from ea_chatbot.graph.nodes.executor import executor_node
-
-@pytest.fixture
-def mock_settings():
-    with patch("ea_chatbot.graph.nodes.executor.Settings") as MockSettings:
-        mock_settings_instance = MagicMock()
-        mock_settings_instance.db_host = "localhost"
-        mock_settings_instance.db_port = 5432
-        mock_settings_instance.db_user = "user"
-        mock_settings_instance.db_pswd = "pass"
-        mock_settings_instance.db_name = "test_db"
-        mock_settings_instance.db_table = "test_table"
-        MockSettings.return_value = mock_settings_instance
-        yield mock_settings_instance
-
-@pytest.fixture
-def mock_db_client():
-    with patch("ea_chatbot.graph.nodes.executor.DBClient") as MockDBClient:
-        mock_client_instance = MagicMock()
-        MockDBClient.return_value = mock_client_instance
-        yield mock_client_instance
-
-def test_executor_node_success_simple_print(mock_settings, mock_db_client):
-    """Test executing simple code that prints to stdout."""
-    state = {
-        "code": "print('Hello, World!')",
-        "question": "test",
-        "messages": []
-    }
-    
-    result = executor_node(state)
-    
-    assert "code_output" in result
-    assert "Hello, World!" in result["code_output"]
-    assert result["error"] is None
-    assert result["plots"] == []
-    assert result["dfs"] == {}
-
-def test_executor_node_success_dataframe(mock_settings, mock_db_client):
-    """Test executing code that creates a DataFrame."""
-    code = """
-import pandas as pd
-df = pd.DataFrame({'a': [1, 2], 'b': [3, 4]})
-print(df)
-"""
-    state = {
-        "code": code,
-        "question": "test",
-        "messages": []
-    }
-    
-    result = executor_node(state)
-    
-    assert "code_output" in result
-    assert "a  b" in result["code_output"] # Check part of DF string representation
-    assert "dfs" in result
-    assert "df" in result["dfs"]
-    assert isinstance(result["dfs"]["df"], pd.DataFrame)
-
-def test_executor_node_success_plot(mock_settings, mock_db_client):
-    """Test executing code that generates a plot."""
-    code = """
-import matplotlib.pyplot as plt
-fig = plt.figure()
-plots.append(fig)
-print('Plot generated')
-"""
-    state = {
-        "code": code,
-        "question": "test",
-        "messages": []
-    }
-    
-    result = executor_node(state)
-    
-    assert "Plot generated" in result["code_output"]
-    assert "plots" in result
-    assert len(result["plots"]) == 1
-    assert isinstance(result["plots"][0], Figure)
-
-def test_executor_node_error_syntax(mock_settings, mock_db_client):
-    """Test executing code with a syntax error."""
-    state = {
-        "code": "print('Hello World", # Missing closing quote
-        "question": "test",
-        "messages": []
-    }
-    
-    result = executor_node(state)
-    
-    assert result["error"] is not None
-    assert "SyntaxError" in result["error"]
-
-def test_executor_node_error_runtime(mock_settings, mock_db_client):
-    """Test executing code with a runtime error."""
-    state = {
-        "code": "print(1 / 0)",
-        "question": "test",
-        "messages": []
-    }
-    
-    result = executor_node(state)
-    
-    assert result["error"] is not None
-    assert "ZeroDivisionError" in result["error"]
-
-def test_executor_node_no_code(mock_settings, mock_db_client):
-    """Test handling when no code is provided."""
-    state = {
-        "code": None,
-        "question": "test",
-        "messages": []
-    }
-    
-    result = executor_node(state)
-    
-    assert "error" in result
-    assert "No code provided" in result["error"]

backend/tests/test_logging_e2e.py

@@ -2,7 +2,7 @@ import json
 import pytest
 import logging
 from unittest.mock import patch
-from ea_chatbot.graph.workflow import app
+from ea_chatbot.graph.workflow import create_workflow
 from ea_chatbot.graph.state import AgentState
 from ea_chatbot.utils.logging import get_logger
 from langchain_community.chat_models import FakeListChatModel
@@ -43,10 +43,10 @@ def test_logging_e2e_json_output(tmp_path):
         "question": "Who won in 2024?",
         "analysis": None,
         "next_action": "",
-        "plan": None,
-        "code": None,
-        "code_output": None,
-        "error": None,
+        "iterations": 0,
+        "checklist": [],
+        "current_step": 0,
+        "vfs": {},
         "plots": [],
         "dfs": {}
     }
@@ -57,6 +57,20 @@ def test_logging_e2e_json_output(tmp_path):
     
     fake_clarify = FakeListChatModel(responses=["Please specify."])
     
+    # Create a test app without interrupts
+    # We need to manually compile it here to avoid the global 'app' which has interrupts
+    from langgraph.graph import StateGraph, END
+    from ea_chatbot.graph.nodes.query_analyzer import query_analyzer_node
+    from ea_chatbot.graph.nodes.clarification import clarification_node
+    
+    workflow = StateGraph(AgentState)
+    workflow.add_node("query_analyzer", query_analyzer_node)
+    workflow.add_node("clarification", clarification_node)
+    workflow.set_entry_point("query_analyzer")
+    workflow.add_edge("query_analyzer", "clarification")
+    workflow.add_edge("clarification", END)
+    test_app = workflow.compile()
+    
     with patch("ea_chatbot.graph.nodes.query_analyzer.get_llm_model") as mock_llm_factory:
         mock_llm_factory.return_value = fake_analyzer
         
@@ -64,7 +78,7 @@ def test_logging_e2e_json_output(tmp_path):
             mock_clarify_llm_factory.return_value = fake_clarify
             
             # Run the graph
-            list(app.stream(initial_state))
+            test_app.invoke(initial_state)
         
     # Verify file content
     assert log_file.exists()

backend/tests/test_multi_turn_planner_researcher.py

@@ -1,24 +1,27 @@
 import pytest
 from unittest.mock import MagicMock, patch
-from langchain_core.messages import HumanMessage, AIMessage
 from ea_chatbot.graph.nodes.planner import planner_node
-from ea_chatbot.graph.nodes.researcher import researcher_node
-from ea_chatbot.graph.nodes.summarizer import summarizer_node
-from ea_chatbot.schemas import TaskPlanResponse
+from ea_chatbot.schemas import ChecklistResponse, ChecklistTask
+from langchain_core.messages import HumanMessage, AIMessage
 
 @pytest.fixture
 def mock_state_with_history():
     return {
         "messages": [
-            HumanMessage(content="Show me the 2024 results for Florida"),
-            AIMessage(content="Here are the results for Florida in 2024...")
+            HumanMessage(content="What about NJ?"),
+            AIMessage(content="NJ has 9 million voters.")
         ],
-        "question": "What about in New Jersey?",
-        "analysis": {"data_required": ["2024 results", "New Jersey"], "unknowns": [], "ambiguities": [], "conditions": []},
+        "question": "Show me the breakdown by county for 2024",
+        "analysis": {
+            "data_required": ["2024 results", "New Jersey"],
+            "unknowns": [],
+            "ambiguities": [],
+            "conditions": []
+        },
         "next_action": "plan",
-        "summary": "The user is asking about 2024 election results.",
-        "plan": "Plan steps...",
-        "code_output": "Code output..."
+        "summary": "The user is asking about NJ 2024 results.",
+        "checklist": [],
+        "current_step": 0
     }
 
 @patch("ea_chatbot.graph.nodes.planner.get_llm_model")
@@ -31,49 +34,17 @@ def test_planner_uses_history_and_summary(mock_prompt, mock_get_summary, mock_ge
     mock_structured_llm = MagicMock()
     mock_llm_instance.with_structured_output.return_value = mock_structured_llm
     
-    mock_structured_llm.invoke.return_value = TaskPlanResponse(
+    mock_structured_llm.invoke.return_value = ChecklistResponse(
         goal="goal",
         reflection="reflection",
-        context={
-            "initial_context": "context",
-            "assumptions": [],
-            "constraints": []
-        },
-        steps=["Step 1: test"]
+        checklist=[ChecklistTask(task="Step 1: test", worker="data_analyst")]
     )
     
     planner_node(mock_state_with_history)
     
-    mock_prompt.format_messages.assert_called_once()
-    kwargs = mock_prompt.format_messages.call_args[1]
-    assert kwargs["question"] == "What about in New Jersey?"
-    assert kwargs["summary"] == mock_state_with_history["summary"]
-    assert len(kwargs["history"]) == 2
-
-@patch("ea_chatbot.graph.nodes.researcher.get_llm_model")
-@patch("ea_chatbot.graph.nodes.researcher.RESEARCHER_PROMPT")
-def test_researcher_uses_history_and_summary(mock_prompt, mock_get_llm, mock_state_with_history):
-    mock_llm_instance = MagicMock()
-    mock_get_llm.return_value = mock_llm_instance
-    
-    researcher_node(mock_state_with_history)
-    
-    mock_prompt.format_messages.assert_called_once()
-    kwargs = mock_prompt.format_messages.call_args[1]
-    assert kwargs["question"] == "What about in New Jersey?"
-    assert kwargs["summary"] == mock_state_with_history["summary"]
-    assert len(kwargs["history"]) == 2
-
-@patch("ea_chatbot.graph.nodes.summarizer.get_llm_model")
-@patch("ea_chatbot.graph.nodes.summarizer.SUMMARIZER_PROMPT")
-def test_summarizer_uses_history_and_summary(mock_prompt, mock_get_llm, mock_state_with_history):
-    mock_llm_instance = MagicMock()
-    mock_get_llm.return_value = mock_llm_instance
-    
-    summarizer_node(mock_state_with_history)
-    
-    mock_prompt.format_messages.assert_called_once()
-    kwargs = mock_prompt.format_messages.call_args[1]
-    assert kwargs["question"] == "What about in New Jersey?"
-    assert kwargs["summary"] == mock_state_with_history["summary"]
-    assert len(kwargs["history"]) == 2
+    # Verify history and summary were passed to prompt format
+    # We check the arguments passed to the mock_prompt.format_messages
+    call_args = mock_prompt.format_messages.call_args[1]
+    assert call_args["summary"] == "The user is asking about NJ 2024 results."
+    assert len(call_args["history"]) == 2
+    assert "breakdown by county" in call_args["question"]

backend/tests/test_node_model_selection.py

@@ -0,0 +1,126 @@
+import pytest
+from unittest.mock import MagicMock, patch
+from ea_chatbot.graph.nodes.query_analyzer import query_analyzer_node
+from ea_chatbot.graph.nodes.planner import planner_node
+from ea_chatbot.graph.nodes.reflector import reflector_node
+from ea_chatbot.graph.nodes.synthesizer import synthesizer_node
+from ea_chatbot.graph.workers.data_analyst.nodes.coder import coder_node as analyst_coder
+from ea_chatbot.graph.workers.researcher.nodes.searcher import searcher_node as researcher_searcher
+from ea_chatbot.graph.state import AgentState
+from ea_chatbot.graph.workers.data_analyst.state import WorkerState as AnalystState
+from ea_chatbot.graph.workers.researcher.state import WorkerState as ResearcherState
+from ea_chatbot.config import Settings
+
+@pytest.fixture
+def mock_settings():
+    settings = Settings()
+    settings.query_analyzer_llm.model = "model-qa"
+    settings.planner_llm.model = "model-planner"
+    settings.reflector_llm.model = "model-reflector"
+    settings.synthesizer_llm.model = "model-synthesizer"
+    settings.coder_llm.model = "model-coder"
+    settings.researcher_llm.model = "model-researcher"
+    return settings
+
+@patch("ea_chatbot.graph.nodes.query_analyzer.get_llm_model")
+@patch("ea_chatbot.graph.nodes.query_analyzer.Settings")
+def test_query_analyzer_model(mock_settings_cls, mock_get_llm, mock_settings):
+    mock_settings_cls.return_value = mock_settings
+    mock_llm = MagicMock()
+    mock_get_llm.return_value = mock_llm
+    
+    state = AgentState(question="test", messages=[], checklist=[], current_step=0, iterations=0, vfs={}, plots=[], dfs={}, next_action="", analysis={})
+    
+    try:
+        query_analyzer_node(state)
+    except:
+        pass # We only care about the call
+    
+    # Verify it was called with the QA config
+    args = mock_get_llm.call_args[0][0]
+    assert args.model == "model-qa"
+
+@patch("ea_chatbot.graph.nodes.planner.get_llm_model")
+@patch("ea_chatbot.graph.nodes.planner.Settings")
+def test_planner_model(mock_settings_cls, mock_get_llm, mock_settings):
+    mock_settings_cls.return_value = mock_settings
+    mock_llm = MagicMock()
+    mock_get_llm.return_value = mock_llm
+    
+    state = AgentState(question="test", messages=[], checklist=[], current_step=0, iterations=0, vfs={}, plots=[], dfs={}, next_action="", analysis={})
+    
+    try:
+        planner_node(state)
+    except:
+        pass
+    
+    args = mock_get_llm.call_args[0][0]
+    assert args.model == "model-planner"
+
+@patch("ea_chatbot.graph.nodes.reflector.get_llm_model")
+@patch("ea_chatbot.graph.nodes.reflector.Settings")
+def test_reflector_model(mock_settings_cls, mock_get_llm, mock_settings):
+    mock_settings_cls.return_value = mock_settings
+    mock_llm = MagicMock()
+    mock_get_llm.return_value = mock_llm
+    
+    state = AgentState(question="test", messages=[], checklist=[{"task": "T1"}], current_step=0, iterations=0, vfs={}, plots=[], dfs={}, next_action="", analysis={}, summary="Worker did X")
+    
+    try:
+        reflector_node(state)
+    except:
+        pass
+    
+    args = mock_get_llm.call_args[0][0]
+    assert args.model == "model-reflector"
+
+@patch("ea_chatbot.graph.nodes.synthesizer.get_llm_model")
+@patch("ea_chatbot.graph.nodes.synthesizer.Settings")
+def test_synthesizer_model(mock_settings_cls, mock_get_llm, mock_settings):
+    mock_settings_cls.return_value = mock_settings
+    mock_llm = MagicMock()
+    mock_get_llm.return_value = mock_llm
+    
+    state = AgentState(question="test", messages=[], checklist=[], current_step=0, iterations=0, vfs={}, plots=[], dfs={}, next_action="", analysis={})
+    
+    try:
+        synthesizer_node(state)
+    except:
+        pass
+    
+    args = mock_get_llm.call_args[0][0]
+    assert args.model == "model-synthesizer"
+
+@patch("ea_chatbot.graph.workers.data_analyst.nodes.coder.get_llm_model")
+@patch("ea_chatbot.graph.workers.data_analyst.nodes.coder.Settings")
+def test_analyst_coder_model(mock_settings_cls, mock_get_llm, mock_settings):
+    mock_settings_cls.return_value = mock_settings
+    mock_llm = MagicMock()
+    mock_get_llm.return_value = mock_llm
+    
+    state = AnalystState(task="test", messages=[], iterations=0, vfs_state={}, plots=[], result=None, code=None, output=None, error=None)
+    
+    try:
+        analyst_coder(state)
+    except:
+        pass
+    
+    args = mock_get_llm.call_args[0][0]
+    assert args.model == "model-coder"
+
+@patch("ea_chatbot.graph.workers.researcher.nodes.searcher.get_llm_model")
+@patch("ea_chatbot.graph.workers.researcher.nodes.searcher.Settings")
+def test_researcher_searcher_model(mock_settings_cls, mock_get_llm, mock_settings):
+    mock_settings_cls.return_value = mock_settings
+    mock_llm = MagicMock()
+    mock_get_llm.return_value = mock_llm
+    
+    state = ResearcherState(task="test", messages=[], iterations=0, result=None, queries=[], raw_results=[])
+    
+    try:
+        researcher_searcher(state)
+    except:
+        pass
+    
+    args = mock_get_llm.call_args[0][0]
+    assert args.model == "model-researcher"

backend/tests/test_orchestrator_delegate.py

@@ -0,0 +1,56 @@
+from ea_chatbot.graph.nodes.delegate import delegate_node
+from ea_chatbot.graph.state import AgentState
+
+def test_delegate_node_data_analyst():
+    """Verify that the delegate node routes to data_analyst."""
+    state = AgentState(
+        checklist=[{"task": "Analyze data", "worker": "data_analyst"}],
+        current_step=0,
+        messages=[],
+        question="test",
+        analysis={},
+        next_action="",
+        iterations=0,
+        vfs={},
+        plots=[],
+        dfs={}
+    )
+    
+    result = delegate_node(state)
+    assert result["next_action"] == "data_analyst"
+
+def test_delegate_node_researcher():
+    """Verify that the delegate node routes to researcher."""
+    state = AgentState(
+        checklist=[{"task": "Search web", "worker": "researcher"}],
+        current_step=0,
+        messages=[],
+        question="test",
+        analysis={},
+        next_action="",
+        iterations=0,
+        vfs={},
+        plots=[],
+        dfs={}
+    )
+    
+    result = delegate_node(state)
+    assert result["next_action"] == "researcher"
+
+def test_delegate_node_finished():
+    """Verify that the delegate node routes to summarize if checklist is complete."""
+    state = AgentState(
+        checklist=[{"task": "Task 1", "worker": "data_analyst"}],
+        current_step=1, # Already finished the only task
+        messages=[],
+        question="test",
+        analysis={},
+        next_action="",
+        iterations=0,
+        vfs={},
+        plots=[],
+        dfs={}
+    )
+    
+    result = delegate_node(state)
+    assert result["next_action"] == "summarize"

backend/tests/test_orchestrator_loop.py

@@ -0,0 +1,136 @@
+import pytest
+from unittest.mock import MagicMock
+from ea_chatbot.graph.workflow import create_workflow
+from ea_chatbot.graph.state import AgentState
+from langchain_core.messages import AIMessage, HumanMessage
+
+def test_orchestrator_full_flow():
+    """Verify the full Orchestrator-Workers flow via direct node injection."""
+    
+    mock_analyzer = MagicMock()
+    mock_planner = MagicMock()
+    mock_delegate = MagicMock()
+    mock_worker = MagicMock()
+    mock_reflector = MagicMock()
+    mock_synthesizer = MagicMock()
+    mock_summarize_conv = MagicMock()
+    
+    # 1. Analyzer: Proceed to planning
+    mock_analyzer.return_value = {"next_action": "plan"}
+    
+    # 2. Planner: Generate checklist
+    mock_planner.return_value = {
+        "checklist": [{"task": "T1", "worker": "data_analyst"}],
+        "current_step": 0
+    }
+    
+    # 3. Delegate: Route to data_analyst
+    mock_delegate.side_effect = [
+        {"next_action": "data_analyst"}, # First call
+        {"next_action": "summarize"}    # Second call (after reflector)
+    ]
+    
+    # 4. Worker: Success
+    mock_worker.return_value = {
+        "messages": [AIMessage(content="Worker result")],
+        "vfs": {"res.txt": "data"}
+    }
+    
+    # 5. Reflector: Advance
+    mock_reflector.return_value = {
+        "current_step": 1,
+        "next_action": "delegate"
+    }
+    
+    # 6. Synthesizer: Final answer
+    mock_synthesizer.return_value = {
+        "messages": [AIMessage(content="Final synthesized answer")],
+        "next_action": "end"
+    }
+    
+    # 7. Summarize Conv: End
+    mock_summarize_conv.return_value = {"summary": "Done"}
+    
+    # Create workflow with injected mocks
+    app = create_workflow(
+        query_analyzer=mock_analyzer,
+        planner=mock_planner,
+        delegate=mock_delegate,
+        data_analyst_worker=mock_worker,
+        reflector=mock_reflector,
+        synthesizer=mock_synthesizer,
+        summarize_conversation=mock_summarize_conv
+    )
+    
+    initial_state = AgentState(
+        messages=[HumanMessage(content="Explain results")],
+        question="Explain results",
+        analysis={},
+        next_action="",
+        iterations=0,
+        checklist=[],
+        current_step=0,
+        vfs={},
+        plots=[],
+        dfs={}
+    )
+    
+    final_state = app.invoke(initial_state)
+    
+    assert mock_analyzer.called
+    assert mock_planner.called
+    assert mock_delegate.call_count == 2
+    assert mock_worker.called
+    assert mock_reflector.called
+    assert mock_synthesizer.called
+    assert "Final synthesized answer" in [m.content for m in final_state["messages"]]
+
+def test_orchestrator_researcher_flow():
+    """Verify that the Orchestrator can route to the researcher worker."""
+    
+    mock_analyzer = MagicMock()
+    mock_planner = MagicMock()
+    mock_delegate = MagicMock()
+    mock_researcher = MagicMock()
+    mock_reflector = MagicMock()
+    mock_synthesizer = MagicMock()
+    
+    mock_analyzer.return_value = {"next_action": "plan"}
+    mock_planner.return_value = {
+        "checklist": [{"task": "Search news", "worker": "researcher"}],
+        "current_step": 0
+    }
+    mock_delegate.side_effect = [
+        {"next_action": "researcher"},
+        {"next_action": "summarize"}
+    ]
+    mock_researcher.return_value = {"messages": [AIMessage(content="News found")]}
+    mock_reflector.return_value = {"current_step": 1, "next_action": "delegate"}
+    mock_synthesizer.return_value = {"messages": [AIMessage(content="Final News Summary")], "next_action": "end"}
+    
+    app = create_workflow(
+        query_analyzer=mock_analyzer,
+        planner=mock_planner,
+        delegate=mock_delegate,
+        researcher_worker=mock_researcher,
+        reflector=mock_reflector,
+        synthesizer=mock_synthesizer
+    )
+    
+    initial_state = AgentState(
+        messages=[HumanMessage(content="What's the news?")],
+        question="What's the news?",
+        analysis={},
+        next_action="",
+        iterations=0,
+        checklist=[],
+        current_step=0,
+        vfs={},
+        plots=[],
+        dfs={}
+    )
+    
+    final_state = app.invoke(initial_state)
+    
+    assert mock_researcher.called
+    assert "Final News Summary" in [m.content for m in final_state["messages"]]

backend/tests/test_orchestrator_planner.py

@@ -0,0 +1,34 @@
+from typing import get_type_hints, List
+from ea_chatbot.graph.nodes.planner import planner_node
+from ea_chatbot.graph.state import AgentState
+
+def test_planner_node_checklist():
+    """Verify that the planner node generates a checklist."""
+    state = AgentState(
+        messages=[],
+        question="How many voters are in Florida and what is the current news?",
+        analysis={"requires_dataset": True},
+        next_action="plan",
+        iterations=0,
+        checklist=[],
+        current_step=0,
+        vfs={},
+        plots=[],
+        dfs={}
+    )
+    
+    # Mocking the LLM would be ideal, but for now we'll check the returned keys
+    # and assume the implementation provides them.
+    # In a real TDD, we'd mock the LLM to return a specific structure.
+    
+    # For now, let's assume the task is to update 'planner_node' to return these keys.
+    result = planner_node(state)
+    
+    assert "checklist" in result
+    assert isinstance(result["checklist"], list)
+    assert len(result["checklist"]) > 0
+    assert "task" in result["checklist"][0]
+    assert "worker" in result["checklist"][0] # 'data_analyst' or 'researcher'
+    
+    assert "current_step" in result
+    assert result["current_step"] == 0

backend/tests/test_orchestrator_reflector.py

@@ -0,0 +1,30 @@
+from unittest.mock import MagicMock, patch
+from ea_chatbot.graph.nodes.reflector import reflector_node
+from ea_chatbot.graph.state import AgentState
+
+def test_reflector_node_satisfied():
+    """Verify that the reflector node increments current_step when satisfied."""
+    state = AgentState(
+        checklist=[{"task": "Analyze votes", "worker": "data_analyst"}],
+        current_step=0,
+        messages=[],
+        question="test",
+        analysis={},
+        next_action="",
+        iterations=0,
+        vfs={},
+        plots=[],
+        dfs={},
+        summary="Worker successfully analyzed votes and found 5 million."
+    )
+    
+    # Mocking the LLM to return 'satisfied=True'
+    with patch("ea_chatbot.graph.nodes.reflector.get_llm_model") as mock_get_llm:
+        mock_llm = MagicMock()
+        mock_llm.with_structured_output.return_value.invoke.return_value = MagicMock(satisfied=True, reasoning="Good.")
+        mock_get_llm.return_value = mock_llm
+        
+        result = reflector_node(state)
+        
+        assert result["current_step"] == 1
+        assert result["next_action"] == "delegate" # Route back to delegate for next task

backend/tests/test_orchestrator_synthesizer.py

@@ -0,0 +1,31 @@
+from unittest.mock import MagicMock, patch
+from ea_chatbot.graph.nodes.synthesizer import synthesizer_node
+from ea_chatbot.graph.state import AgentState
+from langchain_core.messages import AIMessage
+
+def test_synthesizer_node_success():
+    """Verify that the synthesizer node produces a final response."""
+    state = AgentState(
+        messages=[AIMessage(content="Worker 1 found data."), AIMessage(content="Worker 2 searched web.")],
+        question="What are the results?",
+        checklist=[],
+        current_step=0,
+        iterations=0,
+        vfs={},
+        plots=[],
+        dfs={},
+        next_action="",
+        analysis={}
+    )
+    
+    # Mocking the LLM
+    with patch("ea_chatbot.graph.nodes.synthesizer.get_llm_model") as mock_get_llm:
+        mock_llm = MagicMock()
+        mock_llm.invoke.return_value = AIMessage(content="Final synthesized answer.")
+        mock_get_llm.return_value = mock_llm
+        
+        result = synthesizer_node(state)
+        
+        assert len(result["messages"]) == 1
+        assert result["messages"][0].content == "Final synthesized answer."
+        assert result["next_action"] == "end"

backend/tests/test_planner.py

@@ -1,6 +1,7 @@
 import pytest
 from unittest.mock import MagicMock, patch
 from ea_chatbot.graph.nodes.planner import planner_node
+from ea_chatbot.schemas import ChecklistResponse, ChecklistTask
 
 @pytest.fixture
 def mock_state():
@@ -8,39 +9,40 @@ def mock_state():
         "messages": [],
         "question": "Show me results for New Jersey",
         "analysis": {
-            # "requires_dataset" removed as it's no longer used
             "expert": "Data Analyst",
             "data": "NJ data",
             "unknown": "results",
             "condition": "state=NJ"
         },
         "next_action": "plan",
-        "plan": None
+        "checklist": [],
+        "current_step": 0
     }
 
 @patch("ea_chatbot.graph.nodes.planner.get_llm_model")
 @patch("ea_chatbot.utils.database_inspection.get_data_summary")
 def test_planner_node(mock_get_summary, mock_get_llm, mock_state):
-    """Test planner node with unified prompt."""
+    """Test planner node with checklist generation."""
     mock_get_summary.return_value = "Column: Name, Type: text"
     
     mock_llm = MagicMock()
     mock_get_llm.return_value = mock_llm
     
-    from ea_chatbot.schemas import TaskPlanResponse, TaskPlanContext
-    mock_plan = TaskPlanResponse(
+    mock_response = ChecklistResponse(
         goal="Get NJ results",
         reflection="The user wants NJ results",
-        context=TaskPlanContext(initial_context="NJ data", assumptions=[], constraints=[]),
-        steps=["Step 1: Load data", "Step 2: Filter by NJ"]
+        checklist=[
+            ChecklistTask(task="Query NJ data", worker="data_analyst")
+        ]
     )
-    mock_llm.with_structured_output.return_value.invoke.return_value = mock_plan
+    mock_llm.with_structured_output.return_value.invoke.return_value = mock_response
     
     result = planner_node(mock_state)
     
-    assert "plan" in result
-    assert "Step 1: Load data" in result["plan"]
-    assert "Step 2: Filter by NJ" in result["plan"]
+    assert "checklist" in result
+    assert result["checklist"][0]["task"] == "Query NJ data"
+    assert result["current_step"] == 0
+    assert result["summary"] == "The user wants NJ results"
     
     # Verify helper was called
-    mock_get_summary.assert_called_once()
+    mock_get_summary.assert_called_once()

backend/tests/test_researcher.py

@@ -1,34 +0,0 @@
-import pytest
-from unittest.mock import MagicMock, patch
-from langchain_core.messages import AIMessage
-from langchain_openai import ChatOpenAI
-from ea_chatbot.graph.nodes.researcher import researcher_node
-
-@pytest.fixture
-def mock_llm():
-    with patch("ea_chatbot.graph.nodes.researcher.get_llm_model") as mock_get_llm:
-        mock_llm_instance = MagicMock(spec=ChatOpenAI)
-        mock_get_llm.return_value = mock_llm_instance
-        yield mock_llm_instance
-
-def test_researcher_node_success(mock_llm):
-    """Test that researcher_node invokes LLM with web_search tool and returns messages."""
-    state = {
-        "question": "What is the capital of France?",
-        "messages": []
-    }
-    
-    mock_llm_with_tools = MagicMock()
-    mock_llm.bind_tools.return_value = mock_llm_with_tools
-    mock_llm_with_tools.invoke.return_value = AIMessage(content="The capital of France is Paris.")
-    
-    result = researcher_node(state)
-    
-    assert mock_llm.bind_tools.called
-    # Check that it was called with web_search
-    args, kwargs = mock_llm.bind_tools.call_args
-    assert {"type": "web_search"} in args[0]
-    
-    assert mock_llm_with_tools.invoke.called
-    assert "messages" in result
-    assert result["messages"][0].content == "The capital of France is Paris."

backend/tests/test_researcher_search_tools.py

@@ -1,62 +0,0 @@
-import pytest
-from unittest.mock import MagicMock, patch
-from langchain_core.messages import AIMessage
-from langchain_openai import ChatOpenAI
-from langchain_google_genai import ChatGoogleGenerativeAI
-from ea_chatbot.graph.nodes.researcher import researcher_node
-
-@pytest.fixture
-def base_state():
-    return {
-        "question": "Who won the 2024 election?",
-        "messages": [],
-        "summary": ""
-    }
-
-@patch("ea_chatbot.graph.nodes.researcher.get_llm_model")
-def test_researcher_binds_openai_search(mock_get_llm, base_state):
-    """Test that OpenAI LLM binds 'web_search' tool."""
-    mock_llm = MagicMock(spec=ChatOpenAI)
-    mock_get_llm.return_value = mock_llm
-    
-    mock_llm_with_tools = MagicMock()
-    mock_llm.bind_tools.return_value = mock_llm_with_tools
-    mock_llm_with_tools.invoke.return_value = AIMessage(content="OpenAI Search Result")
-    
-    result = researcher_node(base_state)
-    
-    # Verify bind_tools called with correct OpenAI tool
-    mock_llm.bind_tools.assert_called_once_with([{"type": "web_search"}])
-    assert result["messages"][0].content == "OpenAI Search Result"
-
-@patch("ea_chatbot.graph.nodes.researcher.get_llm_model")
-def test_researcher_binds_google_search(mock_get_llm, base_state):
-    """Test that Google LLM binds 'google_search' tool."""
-    mock_llm = MagicMock(spec=ChatGoogleGenerativeAI)
-    mock_get_llm.return_value = mock_llm
-    
-    mock_llm_with_tools = MagicMock()
-    mock_llm.bind_tools.return_value = mock_llm_with_tools
-    mock_llm_with_tools.invoke.return_value = AIMessage(content="Google Search Result")
-    
-    result = researcher_node(base_state)
-    
-    # Verify bind_tools called with correct Google tool
-    mock_llm.bind_tools.assert_called_once_with([{"google_search": {}}])
-    assert result["messages"][0].content == "Google Search Result"
-
-@patch("ea_chatbot.graph.nodes.researcher.get_llm_model")
-def test_researcher_fallback_on_bind_error(mock_get_llm, base_state):
-    """Test that researcher falls back to basic LLM if bind_tools fails."""
-    mock_llm = MagicMock(spec=ChatOpenAI)
-    mock_get_llm.return_value = mock_llm
-    
-    # Simulate bind_tools failing (e.g. model doesn't support it)
-    mock_llm.bind_tools.side_effect = Exception("Not supported")
-    mock_llm.invoke.return_value = AIMessage(content="Basic Result")
-    
-    result = researcher_node(base_state)
-    
-    # Should still succeed using the base LLM
-    assert result["messages"][0].content == "Basic Result"
-    mock_llm.invoke.assert_called_once()

backend/tests/test_researcher_state.py

@@ -0,0 +1,13 @@
+from typing import get_type_hints, List
+from ea_chatbot.graph.workers.researcher.state import WorkerState
+
+def test_researcher_worker_state():
+    """Verify that Researcher WorkerState has the required fields."""
+    hints = get_type_hints(WorkerState)
+    
+    assert "messages" in hints
+    assert "task" in hints
+    assert "queries" in hints
+    assert "raw_results" in hints
+    assert "iterations" in hints
+    assert "result" in hints

backend/tests/test_researcher_worker.py

@@ -0,0 +1,67 @@
+import pytest
+from unittest.mock import MagicMock
+from ea_chatbot.graph.workers.researcher.workflow import create_researcher_worker, WorkerState
+from ea_chatbot.graph.workers.researcher.mapping import prepare_researcher_input, merge_researcher_output
+from ea_chatbot.graph.state import AgentState
+from langchain_core.messages import AIMessage
+
+def test_researcher_worker_flow():
+    """Verify that the Researcher worker flow works as expected."""
+    mock_searcher = MagicMock()
+    mock_summarizer = MagicMock()
+    
+    mock_searcher.return_value = {
+        "raw_results": ["Result A"],
+        "messages": [AIMessage(content="Search result")]
+    }
+    mock_summarizer.return_value = {"result": "Consolidated Summary"}
+    
+    graph = create_researcher_worker(
+        searcher=mock_searcher,
+        summarizer=mock_summarizer
+    )
+    
+    initial_state = WorkerState(
+        messages=[],
+        task="Find governor",
+        queries=[],
+        raw_results=[],
+        iterations=0,
+        result=None
+    )
+    
+    final_state = graph.invoke(initial_state)
+    
+    assert final_state["result"] == "Consolidated Summary"
+    assert mock_searcher.called
+    assert mock_summarizer.called
+
+def test_researcher_mapping():
+    """Verify that we correctly map states for the researcher."""
+    global_state = AgentState(
+        checklist=[{"task": "Search X", "worker": "researcher"}],
+        current_step=0,
+        messages=[],
+        question="test",
+        analysis={},
+        next_action="",
+        iterations=0,
+        vfs={},
+        plots=[],
+        dfs={}
+    )
+    
+    worker_input = prepare_researcher_input(global_state)
+    assert worker_input["task"] == "Search X"
+    
+    worker_output = WorkerState(
+        messages=[],
+        task="Search X",
+        queries=[],
+        raw_results=[],
+        iterations=1,
+        result="Found X"
+    )
+    
+    updates = merge_researcher_output(worker_output)
+    assert updates["messages"][0].content == "Found X"

backend/tests/test_review_fix_clarification.py

@@ -0,0 +1,55 @@
+import pytest
+from unittest.mock import MagicMock, patch
+from ea_chatbot.graph.workflow import create_workflow
+from ea_chatbot.graph.state import AgentState
+from langchain_core.messages import HumanMessage, AIMessage
+
+def test_clarification_flow_immediate_execution():
+    """Verify that an ambiguous query immediately executes the clarification node without interruption."""
+    
+    mock_analyzer = MagicMock()
+    mock_clarification = MagicMock()
+    
+    # 1. Analyzer returns 'clarify'
+    mock_analyzer.return_value = {"next_action": "clarify"}
+    
+    # 2. Clarification node returns a question
+    mock_clarification.return_value = {"messages": [AIMessage(content="What year?")]}
+    
+    # Create workflow without other nodes since they won't be reached
+    # We still need to provide mock planners etc. to create_workflow
+    app = create_workflow(
+        query_analyzer=mock_analyzer,
+        clarification=mock_clarification,
+        planner=MagicMock(),
+        delegate=MagicMock(),
+        data_analyst_worker=MagicMock(),
+        researcher_worker=MagicMock(),
+        reflector=MagicMock(),
+        synthesizer=MagicMock(),
+        summarize_conversation=MagicMock()
+    )
+    
+    initial_state = AgentState(
+        messages=[HumanMessage(content="Who won?")],
+        question="Who won?",
+        analysis={},
+        next_action="",
+        iterations=0,
+        checklist=[],
+        current_step=0,
+        vfs={},
+        plots=[],
+        dfs={}
+    )
+    
+    # Run the graph
+    final_state = app.invoke(initial_state)
+    
+    # Assertions
+    assert mock_analyzer.called
+    assert mock_clarification.called
+    
+    # Verify the state contains the clarification message
+    assert len(final_state["messages"]) > 0
+    assert "What year?" in [m.content for m in final_state["messages"]]

backend/tests/test_review_fix_integration.py

@@ -0,0 +1,48 @@
+import pytest
+from ea_chatbot.graph.workflow import create_workflow, data_analyst_worker_runnable
+from ea_chatbot.graph.state import AgentState
+from unittest.mock import MagicMock, patch
+from langchain_core.messages import HumanMessage, AIMessage
+
+def test_worker_merge_sets_summary_for_reflector(monkeypatch):
+    """Verify that worker node (runnable) sets the 'summary' field for the Reflector."""
+    
+    state = AgentState(
+        messages=[HumanMessage(content="test")],
+        question="test",
+        checklist=[{"task": "Analyze data", "worker": "data_analyst"}],
+        current_step=0,
+        iterations=0,
+        vfs={},
+        plots=[],
+        dfs={},
+        next_action="",
+        analysis={},
+        summary="Initial Planner Summary" # Stale summary
+    )
+    
+    # Create a mock for the invoke method
+    mock_invoke = MagicMock()
+    mock_invoke.return_value = {
+        "summary": "Actual Worker Findings",
+        "messages": [AIMessage(content="Actual Worker Findings")],
+        "vfs": {},
+        "plots": []
+    }
+    
+    # Manually replace the runnable with a mock object that has an invoke method
+    mock_runnable = MagicMock()
+    mock_runnable.invoke = mock_invoke
+    monkeypatch.setattr("ea_chatbot.graph.workflow.data_analyst_worker_runnable", mock_runnable)
+    
+    # Execute via the module reference (which is now mocked)
+    from ea_chatbot.graph.workflow import data_analyst_worker_runnable
+    updates = data_analyst_worker_runnable.invoke(state)
+    
+    # Verify that 'summary' is in updates and has the worker result
+    assert "summary" in updates
+    assert updates["summary"] == "Actual Worker Findings"
+    
+    # When applied to state, it should overwrite the stale summary
+    state.update(updates)
+    assert state["summary"] == "Actual Worker Findings"

backend/tests/test_review_fix_reflector.py

@@ -0,0 +1,35 @@
+import pytest
+from unittest.mock import MagicMock, patch
+from ea_chatbot.graph.nodes.reflector import reflector_node
+from ea_chatbot.graph.state import AgentState
+
+def test_reflector_does_not_advance_on_failure():
+    """Verify that reflector does not increment current_step if not satisfied."""
+    state = AgentState(
+        checklist=[{"task": "Task 1", "worker": "data_analyst"}],
+        current_step=0,
+        messages=[],
+        question="test",
+        analysis={},
+        next_action="",
+        iterations=0,
+        vfs={},
+        plots=[],
+        dfs={},
+        summary="Failed output"
+    )
+    
+    with patch("ea_chatbot.graph.nodes.reflector.get_llm_model") as mock_get_llm:
+        mock_llm = MagicMock()
+        # Mark as NOT satisfied
+        mock_llm.with_structured_output.return_value.invoke.return_value = MagicMock(satisfied=False, reasoning="Incomplete.")
+        mock_get_llm.return_value = mock_llm
+        
+        result = reflector_node(state)
+        
+        # Should NOT increment (therefore not in the updates dict)
+        assert "current_step" not in result
+        # Should increment iterations
+        assert result["iterations"] == 1
+        # Should probably route to planner or retry
+        assert result["next_action"] == "delegate" # Or 'planner' if we want re-planning

backend/tests/test_review_fix_summary.py

@@ -0,0 +1,51 @@
+import pytest
+from ea_chatbot.graph.workers.data_analyst.mapping import merge_worker_output as merge_analyst
+from ea_chatbot.graph.workers.researcher.mapping import merge_researcher_output as merge_researcher
+from ea_chatbot.graph.workers.data_analyst.state import WorkerState as AnalystState
+from ea_chatbot.graph.workers.researcher.state import WorkerState as ResearcherState
+
+def test_analyst_merge_updates_summary():
+    """Verify that analyst merge updates the global summary."""
+    worker_state = AnalystState(
+        result="Actual Worker Result",
+        messages=[],
+        task="test",
+        iterations=1,
+        vfs_state={},
+        plots=[]
+    )
+    
+    updates = merge_analyst(worker_state)
+    assert "summary" in updates
+    assert updates["summary"] == "Actual Worker Result"
+
+def test_researcher_merge_updates_summary():
+    """Verify that researcher merge updates the global summary."""
+    worker_state = ResearcherState(
+        result="Actual Research Result",
+        messages=[],
+        task="test",
+        iterations=1,
+        queries=[],
+        raw_results=[]
+    )
+    
+    updates = merge_researcher(worker_state)
+    assert "summary" in updates
+    assert updates["summary"] == "Actual Research Result"
+
+def test_merge_handles_none_result():
+    """Verify that merge functions handle None results gracefully."""
+    worker_state = AnalystState(
+        result=None,
+        messages=[],
+        task="test",
+        iterations=1,
+        vfs_state={},
+        plots=[]
+    )
+    
+    updates = merge_analyst(worker_state)
+    assert updates["summary"] is not None
+    assert isinstance(updates["messages"][0].content, str)
+    assert len(updates["messages"][0].content) > 0

backend/tests/test_review_fix_unbounded_loop.py

@@ -0,0 +1,85 @@
+import pytest
+from unittest.mock import MagicMock, patch
+from ea_chatbot.graph.workflow import create_workflow
+from ea_chatbot.graph.state import AgentState
+from langchain_core.messages import AIMessage, HumanMessage
+
+def test_orchestrator_loop_retry_budget():
+    """Verify that the orchestrator loop is bounded and terminates after max retries."""
+    
+    mock_analyzer = MagicMock()
+    mock_planner = MagicMock()
+    mock_delegate = MagicMock()
+    mock_worker = MagicMock()
+    mock_reflector = MagicMock()
+    mock_synthesizer = MagicMock()
+    
+    # 1. Analyzer: Proceed to planning
+    mock_analyzer.return_value = {"next_action": "plan"}
+    
+    # 2. Planner: One task
+    mock_planner.return_value = {
+        "checklist": [{"task": "Unsolvable Task", "worker": "data_analyst"}],
+        "current_step": 0,
+        "iterations": 0
+    }
+    
+    # We'll use the REAL delegate and reflector logic to verify the fix
+    # But we mock the LLM calls inside them if necessary.
+    # Actually, it's easier to just mock the node return values but follow the logic.
+    
+    from ea_chatbot.graph.nodes.delegate import delegate_node
+    from ea_chatbot.graph.nodes.reflector import reflector_node
+    
+    # Mocking the LLM inside reflector to always be unsatisfied
+    with patch("ea_chatbot.graph.nodes.reflector.get_llm_model") as mock_get_llm:
+        mock_llm = MagicMock()
+        # Mark as NOT satisfied
+        mock_llm.with_structured_output.return_value.invoke.return_value = MagicMock(satisfied=False, reasoning="Still bad.")
+        mock_get_llm.return_value = mock_llm
+        
+        app = create_workflow(
+            query_analyzer=mock_analyzer,
+            planner=mock_planner,
+            # delegate=delegate_node, # Use real
+            data_analyst_worker=mock_worker,
+            # reflector=reflector_node, # Use real
+            synthesizer=mock_synthesizer
+        )
+        
+        # Mock worker to return something
+        mock_worker.return_value = {"result": "Bad Output", "messages": [AIMessage(content="Bad")]}
+        mock_synthesizer.return_value = {"messages": [AIMessage(content="Failure Summary")], "next_action": "end"}
+        
+        initial_state = AgentState(
+            messages=[HumanMessage(content="test")],
+            question="test",
+            analysis={},
+            next_action="",
+            iterations=0,
+            checklist=[],
+            current_step=0,
+            vfs={},
+            plots=[],
+            dfs={}
+        )
+        
+        # Run the graph. If fix works, it should hit iterations=3 and route to synthesizer.
+        # We use a recursion_limit higher than our retry budget but low enough to fail fast if unbounded.
+        final_state = app.invoke(initial_state, config={"recursion_limit": 20})
+        
+        # Assertions
+        # 1. We tried 3 times (iterations 0, 1, 2) and failed on 3rd.
+        # Wait, delegate routes to summarize when iterations >= 3.
+        # Reflector increments iterations.
+        # Loop: 
+        # Start: it=0
+        # Delegate (it=0) -> Worker -> Reflector (fail, it=1) -> Delegate (it=1)
+        # Delegate (it=1) -> Worker -> Reflector (fail, it=2) -> Delegate (it=2)
+        # Delegate (it=2) -> Worker -> Reflector (fail, it=3) -> Delegate (it=3)
+        # Delegate (it=3) -> Summarize (it=0)
+        
+        assert final_state["iterations"] == 0 # Reset in delegate or handled in synthesizer
+        # Check if we hit the failure summary
+        assert "Failed to complete task" in final_state["summary"]
+        assert mock_worker.call_count == 3

backend/tests/test_review_fix_vfs.py

@@ -0,0 +1,50 @@
+import pytest
+from ea_chatbot.utils.vfs import VFSHelper
+from ea_chatbot.graph.workers.data_analyst.mapping import prepare_worker_input
+from ea_chatbot.graph.state import AgentState
+
+def test_vfs_isolation_deep_copy():
+    """Verify that worker VFS is deep-copied from global state."""
+    global_vfs = {
+        "data.txt": {
+            "content": "original",
+            "metadata": {"tags": ["a"]}
+        }
+    }
+    state = AgentState(
+        checklist=[{"task": "test", "worker": "data_analyst"}],
+        current_step=0,
+        messages=[],
+        question="test",
+        analysis={},
+        next_action="",
+        iterations=0,
+        vfs=global_vfs,
+        plots=[],
+        dfs={}
+    )
+    
+    worker_input = prepare_worker_input(state)
+    worker_vfs = worker_input["vfs_state"]
+    
+    # Mutate worker VFS nested object
+    worker_vfs["data.txt"]["metadata"]["tags"].append("b")
+    
+    # Global VFS should remain unchanged
+    assert global_vfs["data.txt"]["metadata"]["tags"] == ["a"]
+
+def test_vfs_schema_normalization():
+    """Verify that VFSHelper handles inconsistent VFS entries."""
+    vfs = {
+        "raw.txt": "just a string", # Inconsistent with standard Dict[str, Any] entry
+        "valid.txt": {"content": "data", "metadata": {}}
+    }
+    helper = VFSHelper(vfs)
+    
+    # Should not crash during read
+    content, metadata = helper.read("raw.txt")
+    assert content == "just a string"
+    assert metadata == {}
+    
+    # Should not crash during list/other ops if they assume dict
+    assert "raw.txt" in helper.list()

backend/tests/test_state_extensions.py

@@ -0,0 +1,15 @@
+from typing import get_type_hints, List, Dict, Any
+from ea_chatbot.graph.state import AgentState
+
+def test_agent_state_extensions():
+    """Verify that AgentState has the new DeepAgents fields."""
+    hints = get_type_hints(AgentState)
+    
+    assert "checklist" in hints
+    # checklist: List[Dict[str, Any]] (or similar for tasks)
+    
+    assert "current_step" in hints
+    assert hints["current_step"] == int
+    
+    assert "vfs" in hints
+    # vfs: Dict[str, Any]

backend/tests/test_summarizer.py

@@ -1,47 +0,0 @@
-import pytest
-from unittest.mock import MagicMock, patch
-from langchain_core.messages import AIMessage
-from ea_chatbot.graph.nodes.summarizer import summarizer_node
-
-@pytest.fixture
-def mock_llm():
-    with patch("ea_chatbot.graph.nodes.summarizer.get_llm_model") as mock_get_llm:
-        mock_llm_instance = MagicMock()
-        mock_get_llm.return_value = mock_llm_instance
-        yield mock_llm_instance
-
-def test_summarizer_node_success(mock_llm):
-    """Test that summarizer_node invokes LLM with correct inputs and returns messages."""
-    state = {
-        "question": "What is the total count?",
-        "plan": "1. Run query\n2. Sum results",
-        "code_output": "The total is 100",
-        "messages": []
-    }
-    
-    mock_llm.invoke.return_value = AIMessage(content="The final answer is 100.")
-    
-    result = summarizer_node(state)
-    
-    # Verify LLM was called
-    assert mock_llm.invoke.called
-    
-    # Verify result structure
-    assert "messages" in result
-    assert len(result["messages"]) == 1
-    assert isinstance(result["messages"][0], AIMessage)
-    assert result["messages"][0].content == "The final answer is 100."
-
-def test_summarizer_node_empty_state(mock_llm):
-    """Test handling of empty or minimal state."""
-    state = {
-        "question": "Empty?",
-        "messages": []
-    }
-    
-    mock_llm.invoke.return_value = AIMessage(content="No data provided.")
-    
-    result = summarizer_node(state)
-    
-    assert "messages" in result
-    assert result["messages"][0].content == "No data provided."

backend/tests/test_vfs.py

@@ -0,0 +1,49 @@
+import pytest
+from ea_chatbot.utils.vfs import VFSHelper
+
+def test_vfs_read_write():
+    """Verify that we can write and read from the VFS."""
+    vfs = {}
+    helper = VFSHelper(vfs)
+    
+    helper.write("test.txt", "Hello World", metadata={"type": "text"})
+    assert "test.txt" in vfs
+    assert vfs["test.txt"]["content"] == "Hello World"
+    assert vfs["test.txt"]["metadata"]["type"] == "text"
+    
+    content, metadata = helper.read("test.txt")
+    assert content == "Hello World"
+    assert metadata["type"] == "text"
+
+def test_vfs_list():
+    """Verify that we can list files in the VFS."""
+    vfs = {}
+    helper = VFSHelper(vfs)
+    
+    helper.write("a.txt", "content a")
+    helper.write("b.txt", "content b")
+    
+    files = helper.list()
+    assert "a.txt" in files
+    assert "b.txt" in files
+    assert len(files) == 2
+
+def test_vfs_delete():
+    """Verify that we can delete files from the VFS."""
+    vfs = {}
+    helper = VFSHelper(vfs)
+    
+    helper.write("test.txt", "Hello")
+    helper.delete("test.txt")
+    
+    assert "test.txt" not in vfs
+    assert len(helper.list()) == 0
+
+def test_vfs_not_found():
+    """Verify that reading a non-existent file returns None."""
+    vfs = {}
+    helper = VFSHelper(vfs)
+    
+    content, metadata = helper.read("missing.txt")
+    assert content is None
+    assert metadata is None

backend/tests/test_vfs_robustness.py

@@ -0,0 +1,48 @@
+import pytest
+import threading
+from ea_chatbot.utils.vfs import safe_vfs_copy
+
+def test_safe_vfs_copy_success():
+    """Test standard success case."""
+    vfs = {
+        "test.csv": {"content": "data", "metadata": {"type": "csv"}},
+        "num": 42
+    }
+    copied = safe_vfs_copy(vfs)
+    assert copied == vfs
+    assert copied is not vfs
+    assert copied["test.csv"] is not vfs["test.csv"]
+
+def test_safe_vfs_copy_handles_non_copyable():
+    """Test replacing uncopyable objects with error placeholders."""
+    # A threading.Lock is famously uncopyable
+    lock = threading.Lock()
+    
+    vfs = {
+        "safe_file": "important data",
+        "unsafe_lock": lock
+    }
+    
+    copied = safe_vfs_copy(vfs)
+    
+    # Safe one remains
+    assert copied["safe_file"] == "important data"
+    
+    # Unsafe one is REPLACED with an error dict
+    assert isinstance(copied["unsafe_lock"], dict)
+    assert "content" in copied["unsafe_lock"]
+    assert "ERROR" in copied["unsafe_lock"]["content"]
+    assert copied["unsafe_lock"]["metadata"]["type"] == "error"
+    assert "lock" in str(copied["unsafe_lock"]["metadata"]["error"]).lower()
+    
+    # Original is unchanged (it was a lock)
+    assert vfs["unsafe_lock"] is lock
+
+def test_safe_vfs_copy_preserves_nested_copyable():
+    """Test deepcopy still works for complex but copyable objects."""
+    vfs = {
+        "data": {"a": [1, 2, 3], "b": {"c": True}}
+    }
+    copied = safe_vfs_copy(vfs)
+    assert copied["data"]["a"] == [1, 2, 3]
+    assert copied["data"]["a"] is not vfs["data"]["a"]

backend/tests/test_workflow.py

@@ -1,92 +1,80 @@
+import pytest
 from unittest.mock import MagicMock, patch
-from ea_chatbot.graph.workflow import app
-from ea_chatbot.graph.nodes.query_analyzer import QueryAnalysis
-from ea_chatbot.schemas import TaskPlanResponse, TaskPlanContext, CodeGenerationResponse
+from ea_chatbot.graph.workflow import create_workflow
+from ea_chatbot.graph.state import AgentState
+from langchain_core.messages import AIMessage, HumanMessage
 
-from langchain_core.messages import AIMessage
-
-@patch("ea_chatbot.graph.nodes.query_analyzer.get_llm_model")
-@patch("ea_chatbot.graph.nodes.planner.get_llm_model")
-@patch("ea_chatbot.graph.nodes.coder.get_llm_model")
-@patch("ea_chatbot.graph.nodes.summarizer.get_llm_model")
-@patch("ea_chatbot.graph.nodes.researcher.get_llm_model")
-@patch("ea_chatbot.utils.database_inspection.get_data_summary")
-@patch("ea_chatbot.graph.nodes.executor.Settings")
-@patch("ea_chatbot.graph.nodes.executor.DBClient")
-def test_workflow_full_flow(mock_db_client, mock_settings, mock_get_summary, mock_researcher_llm, mock_summarizer_llm, mock_coder_llm, mock_planner_llm, mock_qa_llm):
-    """Test the flow from query_analyzer through planner to coder."""
+def test_workflow_full_flow():
+    """Test the full Orchestrator-Workers flow using node injection."""
     
-    # Mock Settings for Executor
-    mock_settings_instance = MagicMock()
-    mock_settings_instance.db_host = "localhost"
-    mock_settings_instance.db_port = 5432
-    mock_settings_instance.db_user = "user"
-    mock_settings_instance.db_pswd = "pass"
-    mock_settings_instance.db_name = "test_db"
-    mock_settings_instance.db_table = "test_table"
-    mock_settings.return_value = mock_settings_instance
+    mock_analyzer = MagicMock()
+    mock_planner = MagicMock()
+    mock_delegate = MagicMock()
+    mock_worker = MagicMock()
+    mock_reflector = MagicMock()
+    mock_synthesizer = MagicMock()
+    mock_summarize_conv = MagicMock()
     
-    # Mock DBClient
-    mock_client_instance = MagicMock()
-    mock_db_client.return_value = mock_client_instance
+    # 1. Analyzer: Proceed to planning
+    mock_analyzer.return_value = {"next_action": "plan"}
     
-    # 1. Mock Query Analyzer
-    mock_qa_instance = MagicMock()
-    mock_qa_llm.return_value = mock_qa_instance
-    mock_qa_instance.with_structured_output.return_value.invoke.return_value = QueryAnalysis(
-        data_required=["2024 results"], 
-        unknowns=[], 
-        ambiguities=[],
-        conditions=[], 
-        next_action="plan"
+    # 2. Planner: Generate checklist
+    mock_planner.return_value = {
+        "checklist": [{"task": "Step 1", "worker": "data_analyst"}],
+        "current_step": 0
+    }
+    
+    # 3. Delegate: Route to data_analyst
+    mock_delegate.side_effect = [
+        {"next_action": "data_analyst"}, 
+        {"next_action": "summarize"}
+    ]
+    
+    # 4. Worker: Success
+    mock_worker.return_value = {
+        "messages": [AIMessage(content="Worker Summary")],
+        "vfs": {}
+    }
+    
+    # 5. Reflector: Advance
+    mock_reflector.return_value = {
+        "current_step": 1,
+        "next_action": "delegate"
+    }
+    
+    # 6. Synthesizer: Final answer
+    mock_synthesizer.return_value = {
+        "messages": [AIMessage(content="Final Summary")],
+        "next_action": "end"
+    }
+    
+    # 7. Summarize Conv: End
+    mock_summarize_conv.return_value = {"summary": "Done"}
+    
+    app = create_workflow(
+        query_analyzer=mock_analyzer,
+        planner=mock_planner,
+        delegate=mock_delegate,
+        data_analyst_worker=mock_worker,
+        reflector=mock_reflector,
+        synthesizer=mock_synthesizer,
+        summarize_conversation=mock_summarize_conv
     )
     
-    # 2. Mock Planner
-    mock_planner_instance = MagicMock()
-    mock_planner_llm.return_value = mock_planner_instance
-    mock_get_summary.return_value = "Data summary"
-    mock_planner_instance.with_structured_output.return_value.invoke.return_value = TaskPlanResponse(
-        goal="Task Goal",
-        reflection="Reflection",
-        context=TaskPlanContext(initial_context="Ctx", assumptions=[], constraints=[]),
-        steps=["Step 1"]
-    )
-    
-    # 3. Mock Coder
-    mock_coder_instance = MagicMock()
-    mock_coder_llm.return_value = mock_coder_instance
-    mock_coder_instance.with_structured_output.return_value.invoke.return_value = CodeGenerationResponse(
-        code="print('Hello')",
-        explanation="Explanation"
-    )
-
-    # 4. Mock Summarizer
-    mock_summarizer_instance = MagicMock()
-    mock_summarizer_llm.return_value = mock_summarizer_instance
-    mock_summarizer_instance.invoke.return_value = AIMessage(content="Summary")
-
-    # 5. Mock Researcher (just in case)
-    mock_researcher_instance = MagicMock()
-    mock_researcher_llm.return_value = mock_researcher_instance
-    
-    # Initial state
     initial_state = {
-        "messages": [],
-        "question": "Show me the 2024 results",
+        "messages": [HumanMessage(content="Show me results")],
+        "question": "Show me results",
         "analysis": None,
         "next_action": "",
-        "plan": None,
-        "code": None,
-        "error": None,
+        "iterations": 0,
+        "checklist": [],
+        "current_step": 0,
+        "vfs": {},
         "plots": [],
         "dfs": {}
     }
     
-    # Run the graph
-    # We use recursion_limit to avoid infinite loops in placeholders if any
-    result = app.invoke(initial_state, config={"recursion_limit": 10})
+    result = app.invoke(initial_state, config={"recursion_limit": 20})
     
-    assert result["next_action"] == "plan"
-    assert "plan" in result and result["plan"] is not None
-    assert "code" in result and "print('Hello')" in result["code"]
-    assert "analysis" in result
+    assert "Final Summary" in [m.content for m in result["messages"]]
+    assert result["current_step"] == 1

backend/tests/test_workflow_e2e.py

@@ -1,60 +1,55 @@
 import pytest
 from unittest.mock import MagicMock, patch
-from langchain_core.messages import AIMessage
 from ea_chatbot.graph.workflow import app
-from ea_chatbot.graph.nodes.query_analyzer import QueryAnalysis
-from ea_chatbot.schemas import TaskPlanResponse, TaskPlanContext, CodeGenerationResponse
+from ea_chatbot.schemas import QueryAnalysis, ChecklistResponse, ChecklistTask, CodeGenerationResponse
+from ea_chatbot.graph.state import AgentState
+from langchain_core.messages import AIMessage
 
 @pytest.fixture
 def mock_llms():
-    with patch("ea_chatbot.graph.nodes.query_analyzer.get_llm_model") as mock_qa_llm, \
-         patch("ea_chatbot.graph.nodes.planner.get_llm_model") as mock_planner_llm, \
-         patch("ea_chatbot.graph.nodes.coder.get_llm_model") as mock_coder_llm, \
-         patch("ea_chatbot.graph.nodes.summarizer.get_llm_model") as mock_summarizer_llm, \
-         patch("ea_chatbot.graph.nodes.researcher.get_llm_model") as mock_researcher_llm, \
-                      patch("ea_chatbot.graph.nodes.summarize_conversation.get_llm_model") as mock_summary_llm, \
-                      patch("ea_chatbot.utils.database_inspection.get_data_summary") as mock_get_summary:        
-        mock_get_summary.return_value = "Data summary"
-        
-        # Mock summary LLM to return a simple response
-        mock_summary_instance = MagicMock()
-        mock_summary_llm.return_value = mock_summary_instance
-        mock_summary_instance.invoke.return_value = AIMessage(content="Turn summary")
-
+    with patch("ea_chatbot.graph.nodes.query_analyzer.get_llm_model") as mock_qa, \
+         patch("ea_chatbot.graph.nodes.planner.get_llm_model") as mock_planner, \
+         patch("ea_chatbot.graph.workers.data_analyst.nodes.coder.get_llm_model") as mock_coder, \
+         patch("ea_chatbot.graph.workers.data_analyst.nodes.summarizer.get_llm_model") as mock_worker_summarizer, \
+         patch("ea_chatbot.graph.nodes.synthesizer.get_llm_model") as mock_synthesizer, \
+         patch("ea_chatbot.graph.workers.researcher.nodes.searcher.get_llm_model") as mock_researcher, \
+         patch("ea_chatbot.graph.workers.researcher.nodes.summarizer.get_llm_model") as mock_res_summarizer, \
+         patch("ea_chatbot.graph.nodes.reflector.get_llm_model") as mock_reflector:
         yield {
-            "qa": mock_qa_llm,
-            "planner": mock_planner_llm,
-            "coder": mock_coder_llm,
-            "summarizer": mock_summarizer_llm,
-            "researcher": mock_researcher_llm,
-            "summary": mock_summary_llm
+            "qa": mock_qa,
+            "planner": mock_planner,
+            "coder": mock_coder,
+            "worker_summarizer": mock_worker_summarizer,
+            "synthesizer": mock_synthesizer,
+            "researcher": mock_researcher,
+            "res_summarizer": mock_res_summarizer,
+            "reflector": mock_reflector
         }
 
 def test_workflow_data_analysis_flow(mock_llms):
-    """Test full flow: QueryAnalyzer -> Planner -> Coder -> Executor -> Summarizer."""
+    """Test full flow: QueryAnalyzer -> Planner -> Delegate -> DataAnalyst -> Reflector -> Synthesizer."""
     
-    # 1. Mock Query Analyzer (routes to plan)
+    # 1. Mock Query Analyzer
     mock_qa_instance = MagicMock()
     mock_llms["qa"].return_value = mock_qa_instance
     mock_qa_instance.with_structured_output.return_value.invoke.return_value = QueryAnalysis(
-        data_required=["2024 results"], 
-        unknowns=[], 
+        data_required=["2024 results"],
+        unknowns=[],
         ambiguities=[],
-        conditions=[], 
+        conditions=[],
         next_action="plan"
     )
     
     # 2. Mock Planner
     mock_planner_instance = MagicMock()
     mock_llms["planner"].return_value = mock_planner_instance
-    mock_planner_instance.with_structured_output.return_value.invoke.return_value = TaskPlanResponse(
+    mock_planner_instance.with_structured_output.return_value.invoke.return_value = ChecklistResponse(
         goal="Get results",
         reflection="Reflect",
-        context=TaskPlanContext(initial_context="Ctx", assumptions=[], constraints=[]),
-        steps=["Step 1"]
+        checklist=[ChecklistTask(task="Query Data", worker="data_analyst")]
     )
     
-    # 3. Mock Coder
+    # 3. Mock Coder (Worker)
     mock_coder_instance = MagicMock()
     mock_llms["coder"].return_value = mock_coder_instance
     mock_coder_instance.with_structured_output.return_value.invoke.return_value = CodeGenerationResponse(
@@ -62,10 +57,20 @@ def test_workflow_data_analysis_flow(mock_llms):
         explanation="Explain"
     )
     
-    # 4. Mock Summarizer
-    mock_summarizer_instance = MagicMock()
-    mock_llms["summarizer"].return_value = mock_summarizer_instance
-    mock_summarizer_instance.invoke.return_value = AIMessage(content="Final Summary: Success")
+    # 4. Mock Worker Summarizer
+    mock_ws_instance = MagicMock()
+    mock_llms["worker_summarizer"].return_value = mock_ws_instance
+    mock_ws_instance.invoke.return_value = AIMessage(content="Worker Summary")
+    
+    # 5. Mock Reflector
+    mock_reflector_instance = MagicMock()
+    mock_llms["reflector"].return_value = mock_reflector_instance
+    mock_reflector_instance.with_structured_output.return_value.invoke.return_value = MagicMock(satisfied=True, reasoning="Good.")
+
+    # 6. Mock Synthesizer
+    mock_syn_instance = MagicMock()
+    mock_llms["synthesizer"].return_value = mock_syn_instance
+    mock_syn_instance.invoke.return_value = AIMessage(content="Final Summary: Success")
     
     # Initial state
     initial_state = {
@@ -73,66 +78,77 @@ def test_workflow_data_analysis_flow(mock_llms):
         "question": "Show me 2024 results",
         "analysis": None,
         "next_action": "",
-        "plan": None,
-        "code": None,
-        "error": None,
+        "iterations": 0,
+        "checklist": [],
+        "current_step": 0,
+        "vfs": {},
         "plots": [],
         "dfs": {}
     }
     
     # Run the graph
-    result = app.invoke(initial_state, config={"recursion_limit": 15})
+    result = app.invoke(initial_state, config={"recursion_limit": 20})
     
-    assert result["next_action"] == "plan"
-    assert "Execution Success" in result["code_output"]
-    assert "Final Summary: Success" in result["messages"][-1].content
+    assert "Final Summary: Success" in [m.content for m in result["messages"]]
+    assert result["current_step"] == 1
 
 def test_workflow_research_flow(mock_llms):
-    """Test flow: QueryAnalyzer -> Researcher -> Summarizer."""
+    """Test flow with research task."""
     
-    # 1. Mock Query Analyzer (routes to research)
+    # 1. Mock Query Analyzer
     mock_qa_instance = MagicMock()
     mock_llms["qa"].return_value = mock_qa_instance
     mock_qa_instance.with_structured_output.return_value.invoke.return_value = QueryAnalysis(
-        data_required=[], 
-        unknowns=[], 
+        data_required=[],
+        unknowns=[],
         ambiguities=[],
-        conditions=[], 
+        conditions=[],
         next_action="research"
     )
     
-    # 2. Mock Researcher
-    mock_researcher_instance = MagicMock()
-    mock_llms["researcher"].return_value = mock_researcher_instance
-    # Researcher node uses bind_tools if it's ChatOpenAI/ChatGoogleGenerativeAI
-    # Since it's a MagicMock, it will fallback to using the base instance
-    mock_researcher_instance.invoke.return_value = AIMessage(content="Research Results")
+    # 2. Mock Planner
+    mock_planner_instance = MagicMock()
+    mock_llms["planner"].return_value = mock_planner_instance
+    mock_planner_instance.with_structured_output.return_value.invoke.return_value = ChecklistResponse(
+        goal="Search",
+        reflection="Reflect",
+        checklist=[ChecklistTask(task="Search Web", worker="researcher")]
+    )
     
-    # Also mock bind_tools just in case we ever use spec
-    mock_llm_with_tools = MagicMock()
-    mock_researcher_instance.bind_tools.return_value = mock_llm_with_tools
-    mock_llm_with_tools.invoke.return_value = AIMessage(content="Research Results")
+    # 3. Mock Researcher Searcher
+    mock_res_instance = MagicMock()
+    mock_llms["researcher"].return_value = mock_res_instance
+    mock_res_instance.invoke.return_value = AIMessage(content="Research Result")
     
-    # 3. Mock Summarizer (not used in this flow, but kept for completeness)
-    mock_summarizer_instance = MagicMock()
-    mock_llms["summarizer"].return_value = mock_summarizer_instance
-    mock_summarizer_instance.invoke.return_value = AIMessage(content="Final Summary: Research Success")
+    # 4. Mock Researcher Summarizer
+    mock_rs_instance = MagicMock()
+    mock_llms["res_summarizer"].return_value = mock_rs_instance
+    mock_rs_instance.invoke.return_value = AIMessage(content="Researcher Summary")
+
+    # 5. Mock Reflector
+    mock_reflector_instance = MagicMock()
+    mock_llms["reflector"].return_value = mock_reflector_instance
+    mock_reflector_instance.with_structured_output.return_value.invoke.return_value = MagicMock(satisfied=True, reasoning="Good.")
+
+    # 6. Mock Synthesizer
+    mock_syn_instance = MagicMock()
+    mock_llms["synthesizer"].return_value = mock_syn_instance
+    mock_syn_instance.invoke.return_value = AIMessage(content="Final Research Summary")
     
-    # Initial state
     initial_state = {
         "messages": [],
-        "question": "Who is the governor of Florida?",
+        "question": "Who is the governor?",
         "analysis": None,
         "next_action": "",
-        "plan": None,
-        "code": None,
-        "error": None,
+        "iterations": 0,
+        "checklist": [],
+        "current_step": 0,
+        "vfs": {},
         "plots": [],
         "dfs": {}
     }
     
-    # Run the graph
-    result = app.invoke(initial_state, config={"recursion_limit": 10})
+    result = app.invoke(initial_state, config={"recursion_limit": 20})
     
-    assert result["next_action"] == "research"
-    assert "Research Results" in result["messages"][-1].content
+    assert "Final Research Summary" in [m.content for m in result["messages"]]
+    assert result["current_step"] == 1

frontend/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "frontend",
-  "version": "0.0.0",
+  "version": "0.1.0",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "frontend",
-      "version": "0.0.0",
+      "version": "0.1.0",
       "dependencies": {
         "@hookform/resolvers": "^5.2.2",
         "@tailwindcss/vite": "^4.1.18",

frontend/src/components/chat/ExecutionStatus.tsx

@@ -9,9 +9,9 @@ interface ExecutionStatusProps {
 
 const PHASE_CONFIG = [
   { label: "Analyzing query...", match: "Query analysis complete." },
-  { label: "Generating strategic plan...", match: "Strategic plan generated." },
-  { label: "Writing analysis code...", match: "Analysis code generated." },
-  { label: "Performing data analysis...", match: "Data analysis and visualization complete." }
+  { label: "Generating high-level plan...", match: "Checklist generated." },
+  { label: "Delegating to specialists...", match: "Task assigned." },
+  { label: "Synthesizing final answer...", match: "Final synthesis complete." }
 ]
 
 export function ExecutionStatus({ steps, isComplete, className }: ExecutionStatusProps) {

frontend/src/components/ui/button-variants.ts

@@ -1,7 +1,7 @@
 import { cva } from "class-variance-authority"
 
 export const buttonVariants = cva(
-  "inline-flex items-center justify-center gap-2 whitespace-nowrap rounded-md text-sm font-medium transition-all disabled:pointer-events-none disabled:opacity-50 [&_svg]:pointer-events-none [&_svg:not([class*='size-'])]:size-4 shrink-0 [&_svg]:shrink-0 outline-none focus-visible:border-ring focus-visible:ring-ring/50 focus-visible:ring-[3px] aria-invalid:ring-destructive/20 dark:aria-invalid:ring-destructive/40 aria-invalid:border-destructive",
+  "inline-flex items-center justify-center gap-2 whitespace-nowrap rounded-md text-sm font-medium transition-all cursor-pointer disabled:pointer-events-none disabled:opacity-50 disabled:cursor-default [&_svg]:pointer-events-none [&_svg:not([class*='size-'])]:size-4 shrink-0 [&_svg]:shrink-0 outline-none focus-visible:border-ring focus-visible:ring-ring/50 focus-visible:ring-[3px] aria-invalid:ring-destructive/20 dark:aria-invalid:ring-destructive/40 aria-invalid:border-destructive",
   {
     variants: {
       variant: {

frontend/src/components/ui/button.test.tsx

@@ -0,0 +1,29 @@
+import { render, screen } from "@testing-library/react"
+import { describe, expect, it } from "vitest"
+import { Button } from "./button"
+
+describe("Button component", () => {
+  it("renders with a pointer cursor", () => {
+    render(<Button>Click me</Button>)
+    const button = screen.getByRole("button", { name: /click me/i })
+    expect(button).toHaveClass("cursor-pointer")
+  })
+
+  it("renders different variants with a pointer cursor", () => {
+    const { rerender } = render(<Button variant="outline">Outline</Button>)
+    let button = screen.getByRole("button", { name: /outline/i })
+    expect(button).toHaveClass("cursor-pointer")
+
+    rerender(<Button variant="destructive">Destructive</Button>)
+    button = screen.getByRole("button", { name: /destructive/i })
+    expect(button).toHaveClass("cursor-pointer")
+  })
+
+  it("renders with a default cursor when disabled", () => {
+    render(<Button disabled>Disabled Button</Button>)
+    const button = screen.getByRole("button", { name: /disabled button/i })
+    expect(button).toBeDisabled()
+    expect(button).toHaveClass("disabled:pointer-events-none")
+    expect(button).toHaveClass("disabled:cursor-default")
+  })
+})

frontend/src/components/ui/dropdown-menu.test.tsx

@@ -0,0 +1,74 @@
+import { render, screen } from "@testing-library/react"
+import { describe, expect, it } from "vitest"
+import { 
+  DropdownMenu, 
+  DropdownMenuContent, 
+  DropdownMenuItem, 
+  DropdownMenuSub,
+  DropdownMenuSubTrigger,
+  DropdownMenuSubContent,
+  DropdownMenuPortal
+} from "./dropdown-menu"
+
+describe("DropdownMenu components", () => {
+  it("DropdownMenuItem should have cursor-pointer when enabled", () => {
+    render(
+      <DropdownMenu open={true}>
+        <DropdownMenuContent>
+          <DropdownMenuItem>Enabled Item</DropdownMenuItem>
+        </DropdownMenuContent>
+      </DropdownMenu>
+    )
+    const item = screen.getByText("Enabled Item")
+    expect(item).toHaveClass("cursor-pointer")
+  })
+
+  it("DropdownMenuSubTrigger should have cursor-pointer when enabled", () => {
+    render(
+      <DropdownMenu open={true}>
+        <DropdownMenuContent>
+          <DropdownMenuSub>
+            <DropdownMenuSubTrigger>Sub Trigger</DropdownMenuSubTrigger>
+            <DropdownMenuPortal>
+              <DropdownMenuSubContent />
+            </DropdownMenuPortal>
+          </DropdownMenuSub>
+        </DropdownMenuContent>
+      </DropdownMenu>
+    )
+    const trigger = screen.getByText("Sub Trigger")
+    expect(trigger).toHaveClass("cursor-pointer")
+  })
+
+  it("DropdownMenuItem should show default cursor when disabled", () => {
+    render(
+      <DropdownMenu open={true}>
+        <DropdownMenuContent>
+          <DropdownMenuItem disabled>Disabled Item</DropdownMenuItem>
+        </DropdownMenuContent>
+      </DropdownMenu>
+    )
+    const item = screen.getByText("Disabled Item")
+    // Radix sets data-disabled attribute
+    expect(item).toHaveAttribute("data-disabled")
+    expect(item).toHaveClass("data-[disabled]:cursor-default")
+  })
+
+  it("DropdownMenuSubTrigger should show default cursor when disabled", () => {
+    render(
+      <DropdownMenu open={true}>
+        <DropdownMenuContent>
+          <DropdownMenuSub>
+            <DropdownMenuSubTrigger disabled>Disabled Sub Trigger</DropdownMenuSubTrigger>
+            <DropdownMenuPortal>
+              <DropdownMenuSubContent />
+            </DropdownMenuPortal>
+          </DropdownMenuSub>
+        </DropdownMenuContent>
+      </DropdownMenu>
+    )
+    const trigger = screen.getByText("Disabled Sub Trigger")
+    expect(trigger).toHaveAttribute("data-disabled")
+    expect(trigger).toHaveClass("data-[disabled]:cursor-default")
+  })
+})

frontend/src/components/ui/dropdown-menu.tsx

@@ -27,7 +27,7 @@ const DropdownMenuSubTrigger = React.forwardRef<
   <DropdownMenuPrimitive.SubTrigger
     ref={ref}
     className={cn(
-      "focus:bg-accent data-[state=open]:bg-accent flex cursor-default items-center rounded-sm px-2 py-1.5 text-sm outline-hidden select-none",
+      "focus:bg-accent data-[state=open]:bg-accent flex cursor-pointer items-center rounded-sm px-2 py-1.5 text-sm outline-hidden select-none data-[disabled]:cursor-default data-[disabled]:pointer-events-none data-[disabled]:opacity-50",
       inset && "pl-8",
       className
     )}
@@ -81,7 +81,7 @@ const DropdownMenuItem = React.forwardRef<
   <DropdownMenuPrimitive.Item
     ref={ref}
     className={cn(
-      "focus:bg-accent focus:text-accent-foreground relative flex cursor-default items-center rounded-sm px-2 py-1.5 text-sm outline-hidden transition-colors data-[disabled]:pointer-events-none data-[disabled]:opacity-50 select-none",
+      "focus:bg-accent focus:text-accent-foreground relative flex cursor-pointer items-center rounded-sm px-2 py-1.5 text-sm outline-hidden transition-colors data-[disabled]:pointer-events-none data-[disabled]:opacity-50 data-[disabled]:cursor-default select-none",
       inset && "pl-8",
       className
     )}
@@ -99,7 +99,7 @@ const DropdownMenuCheckboxItem = React.forwardRef<
   <DropdownMenuPrimitive.CheckboxItem
     ref={ref}
     className={cn(
-      "focus:bg-accent focus:text-accent-foreground relative flex cursor-default items-center rounded-sm py-1.5 pr-2 pl-8 text-sm outline-hidden transition-colors data-[disabled]:pointer-events-none data-[disabled]:opacity-50 select-none",
+      "focus:bg-accent focus:text-accent-foreground relative flex cursor-pointer items-center rounded-sm py-1.5 pr-2 pl-8 text-sm outline-hidden transition-colors data-[disabled]:pointer-events-none data-[disabled]:opacity-50 data-[disabled]:cursor-default select-none",
       className
     )}
     checked={checked}
@@ -122,7 +122,7 @@ const DropdownMenuRadioItem = React.forwardRef<
   <DropdownMenuPrimitive.RadioItem
     ref={ref}
     className={cn(
-      "focus:bg-accent focus:text-accent-foreground relative flex cursor-default items-center rounded-sm py-1.5 pr-2 pl-8 text-sm outline-hidden transition-colors data-[disabled]:pointer-events-none data-[disabled]:opacity-50 select-none",
+      "focus:bg-accent focus:text-accent-foreground relative flex cursor-pointer items-center rounded-sm py-1.5 pr-2 pl-8 text-sm outline-hidden transition-colors data-[disabled]:pointer-events-none data-[disabled]:opacity-50 data-[disabled]:cursor-default select-none",
       className
     )}
     {...props}

frontend/src/components/ui/sidebar.test.tsx

@@ -0,0 +1,83 @@
+import { render, screen } from "@testing-library/react"
+import { describe, expect, it } from "vitest"
+import { 
+  SidebarProvider, 
+  SidebarMenuButton, 
+  SidebarMenuAction, 
+  SidebarGroupAction,
+  SidebarMenu,
+  SidebarMenuItem,
+  SidebarGroup
+} from "./sidebar"
+
+describe("Sidebar components", () => {
+  it("SidebarMenuButton should have cursor-pointer", () => {
+    render(
+      <SidebarProvider>
+        <SidebarMenu>
+          <SidebarMenuItem>
+            <SidebarMenuButton>Menu Button</SidebarMenuButton>
+          </SidebarMenuItem>
+        </SidebarMenu>
+      </SidebarProvider>
+    )
+    const button = screen.getByRole("button", { name: /menu button/i })
+    expect(button).toHaveClass("cursor-pointer")
+  })
+
+  it("SidebarMenuAction should have cursor-pointer", () => {
+    render(
+      <SidebarProvider>
+        <SidebarMenu>
+          <SidebarMenuItem>
+            <SidebarMenuButton>Button</SidebarMenuButton>
+            <SidebarMenuAction>Action</SidebarMenuAction>
+          </SidebarMenuItem>
+        </SidebarMenu>
+      </SidebarProvider>
+    )
+    const action = screen.getByRole("button", { name: /action/i })
+    expect(action).toHaveClass("cursor-pointer")
+  })
+
+  it("SidebarGroupAction should have cursor-pointer", () => {
+    render(
+      <SidebarProvider>
+        <SidebarGroup>
+          <SidebarGroupAction>Group Action</SidebarGroupAction>
+        </SidebarGroup>
+      </SidebarProvider>
+    )
+    const action = screen.getByRole("button", { name: /group action/i })
+    expect(action).toHaveClass("cursor-pointer")
+  })
+
+  it("SidebarGroupAction should have cursor-default when disabled", () => {
+    render(
+      <SidebarProvider>
+        <SidebarGroup>
+          <SidebarGroupAction disabled>Disabled Group Action</SidebarGroupAction>
+        </SidebarGroup>
+      </SidebarProvider>
+    )
+    const action = screen.getByRole("button", { name: /disabled group action/i })
+    expect(action).toHaveClass("disabled:cursor-default")
+    expect(action).toBeDisabled()
+  })
+
+  it("SidebarMenuAction should not show pointer when disabled", () => {
+    render(
+      <SidebarProvider>
+        <SidebarMenu>
+          <SidebarMenuItem>
+            <SidebarMenuButton>Button</SidebarMenuButton>
+            <SidebarMenuAction disabled>Disabled Action</SidebarMenuAction>
+          </SidebarMenuItem>
+        </SidebarMenu>
+      </SidebarProvider>
+    )
+    const action = screen.getByRole("button", { name: /disabled action/i })
+    expect(action).toHaveClass("disabled:cursor-default")
+    expect(action).toBeDisabled()
+  })
+})

frontend/src/components/ui/sidebar.tsx

@@ -405,7 +405,7 @@ function SidebarGroupAction({
       data-slot="sidebar-group-action"
       data-sidebar="group-action"
       className={cn(
-        "text-sidebar-foreground ring-sidebar-ring hover:bg-sidebar-accent hover:text-sidebar-accent-foreground absolute top-3.5 right-3 flex aspect-square w-5 items-center justify-center rounded-md p-0 outline-hidden transition-transform focus-visible:ring-2 [&>svg]:size-4 [&>svg]:shrink-0",
+        "text-sidebar-foreground ring-sidebar-ring hover:bg-sidebar-accent hover:text-sidebar-accent-foreground absolute top-3.5 right-3 flex aspect-square w-5 items-center justify-center rounded-md p-0 outline-hidden transition-transform focus-visible:ring-2 [&>svg]:size-4 [&>svg]:shrink-0 cursor-pointer disabled:cursor-default disabled:pointer-events-none disabled:opacity-50",
         // Increases the hit area of the button on mobile.
         "after:absolute after:-inset-2 md:after:hidden",
         "group-data-[collapsible=icon]:hidden",
@@ -477,7 +477,7 @@ function SidebarMenuButton({
       data-size={size}
       data-active={isActive}
       className={cn(
-        "peer/menu-button flex w-full items-center gap-2 overflow-hidden rounded-md p-2 text-left text-sm outline-hidden ring-sidebar-ring transition-[width,height,padding] hover:bg-sidebar-accent hover:text-sidebar-accent-foreground focus-visible:ring-2 active:bg-sidebar-accent active:text-sidebar-accent-foreground disabled:pointer-events-none disabled:opacity-50 group-has-data-[sidebar=menu-action]/menu-item:pr-8 aria-disabled:pointer-events-none aria-disabled:opacity-50 data-[active=true]:bg-sidebar-accent data-[active=true]:font-medium data-[active=true]:text-sidebar-accent-foreground data-[state=open]:hover:bg-sidebar-accent data-[state=open]:hover:text-sidebar-accent-foreground group-data-[collapsible=icon]:size-8! group-data-[collapsible=icon]:p-2! [&>span:last-child]:truncate [&>svg]:size-4 [&>svg]:shrink-0",
+        "peer/menu-button flex w-full items-center gap-2 overflow-hidden rounded-md p-2 text-left text-sm outline-hidden ring-sidebar-ring transition-[width,height,padding] hover:bg-sidebar-accent hover:text-sidebar-accent-foreground focus-visible:ring-2 active:bg-sidebar-accent active:text-sidebar-accent-foreground disabled:pointer-events-none disabled:opacity-50 group-has-data-[sidebar=menu-action]/menu-item:pr-8 aria-disabled:pointer-events-none aria-disabled:opacity-50 data-[active=true]:bg-sidebar-accent data-[active=true]:font-medium data-[active=true]:text-sidebar-accent-foreground data-[state=open]:hover:bg-sidebar-accent data-[state=open]:hover:text-sidebar-accent-foreground group-data-[collapsible=icon]:size-8! group-data-[collapsible=icon]:p-2! [&>span:last-child]:truncate [&>svg]:size-4 [&>svg]:shrink-0 cursor-pointer disabled:cursor-default",
         variant === "default" && "hover:bg-sidebar-accent hover:text-sidebar-accent-foreground",
         variant === "outline" && "bg-background shadow-[0_0_0_1px_hsl(var(--sidebar-border))] hover:bg-sidebar-accent hover:text-sidebar-accent-foreground hover:shadow-[0_0_0_1px_hsl(var(--sidebar-accent))]",
         size === "default" && "h-8 text-sm",
@@ -528,7 +528,7 @@ function SidebarMenuAction({
       data-slot="sidebar-menu-action"
       data-sidebar="menu-action"
       className={cn(
-        "text-sidebar-foreground ring-sidebar-ring hover:bg-sidebar-accent hover:text-sidebar-accent-foreground peer-hover/menu-button:text-sidebar-accent-foreground absolute top-1.5 right-1 flex aspect-square w-5 items-center justify-center rounded-md p-0 outline-hidden transition-transform focus-visible:ring-2 [&>svg]:size-4 [&>svg]:shrink-0",
+        "text-sidebar-foreground ring-sidebar-ring hover:bg-sidebar-accent hover:text-sidebar-accent-foreground peer-hover/menu-button:text-sidebar-accent-foreground absolute top-1.5 right-1 flex aspect-square w-5 items-center justify-center rounded-md p-0 outline-hidden transition-transform focus-visible:ring-2 [&>svg]:size-4 [&>svg]:shrink-0 cursor-pointer disabled:cursor-default disabled:pointer-events-none disabled:opacity-50",
         // Increases the hit area of the button on mobile.
         "after:absolute after:-inset-2 md:after:hidden",
         "peer-data-[size=sm]/menu-button:top-1",

frontend/src/services/chat.test.ts

@@ -3,21 +3,21 @@ import { ChatService, type ChatEvent, type MessageResponse } from "./chat"
 
 describe("ChatService SSE Parsing", () => {
   it("should correctly parse a text stream chunk", () => {
-    const rawChunk = `data: {"type": "on_chat_model_stream", "name": "summarizer", "data": {"chunk": "Hello"}}\n\n`
+    const rawChunk = `data: {"type": "on_chat_model_stream", "name": "synthesizer", "data": {"chunk": "Hello"}}\n\n`
     const events = ChatService.parseSSEChunk(rawChunk)
     
     expect(events).toHaveLength(1)
     expect(events[0]).toEqual({
       type: "on_chat_model_stream",
-      name: "summarizer",
+      name: "synthesizer",
       data: { chunk: "Hello" }
     })
   })
 
   it("should handle multiple events in one chunk", () => {
     const rawChunk = 
-      `data: {"type": "on_chat_model_stream", "name": "summarizer", "data": {"chunk": "Hello"}}\n\n` +
-      `data: {"type": "on_chat_model_stream", "name": "summarizer", "data": {"chunk": " World"}}\n\n`
+      `data: {"type": "on_chat_model_stream", "name": "synthesizer", "data": {"chunk": "Hello"}}\n\n` +
+      `data: {"type": "on_chat_model_stream", "name": "synthesizer", "data": {"chunk": " World"}}\n\n`
     
     const events = ChatService.parseSSEChunk(rawChunk)
     
@@ -25,8 +25,8 @@ describe("ChatService SSE Parsing", () => {
     expect(events[1].data!.chunk).toBe(" World")
   })
 
-  it("should parse encoded plots from executor node", () => {
-    const rawChunk = `data: {"type": "on_chain_end", "name": "executor", "data": {"encoded_plots": ["base64data"]}}\n\n`
+  it("should parse encoded plots from data_analyst_worker node", () => {
+    const rawChunk = `data: {"type": "on_chain_end", "name": "data_analyst_worker", "data": {"encoded_plots": ["base64data"]}}\n\n`
     const events = ChatService.parseSSEChunk(rawChunk)
     
     expect(events[0].data!.encoded_plots).toEqual(["base64data"])
@@ -45,7 +45,7 @@ describe("ChatService Message State Management", () => {
     const messages: MessageResponse[] = [{ id: "1", role: "assistant", content: "Initial", created_at: new Date().toISOString() }]
     const event: ChatEvent = {
       type: "on_chat_model_stream",
-      node: "summarizer",
+      node: "synthesizer",
       data: { chunk: { content: " text" } }
     }
     
@@ -57,7 +57,7 @@ describe("ChatService Message State Management", () => {
     const messages: MessageResponse[] = [{ id: "1", role: "assistant", content: "Analysis", created_at: new Date().toISOString(), plots: [] }]
     const event: ChatEvent = {
       type: "on_chain_end",
-      name: "executor",
+      name: "data_analyst_worker",
       data: { encoded_plots: ["plot1"] }
     }
     

frontend/src/services/chat.ts

@@ -86,7 +86,8 @@ export const ChatService = {
     const { type, name, node, data } = event
     
     // 1. Handle incremental LLM chunks for terminal nodes
-    if (type === "on_chat_model_stream" && (node === "summarizer" || node === "researcher" || node === "clarification")) {
+    // Now using 'synthesizer' for the final user response
+    if (type === "on_chat_model_stream" && (node === "synthesizer" || node === "clarification")) {
       const chunk = data?.chunk?.content || ""
       if (!chunk) return messages
       
@@ -110,7 +111,7 @@ export const ChatService = {
       if (!lastMsg || lastMsg.role !== "assistant") return messages
 
       // Terminal nodes final text
-      if (name === "summarizer" || name === "researcher" || name === "clarification") {
+      if (name === "synthesizer" || name === "clarification") {
         const messages_list = data?.output?.messages
         const msg = messages_list ? messages_list[messages_list.length - 1]?.content : null
         
@@ -121,8 +122,8 @@ export const ChatService = {
         }
       }
 
-      // Plots from executor
-      if (name === "executor" && data?.encoded_plots) {
+      // Plots from data analyst worker
+      if (name === "data_analyst_worker" && data?.encoded_plots) {
         lastMsg.plots = [...(lastMsg.plots || []), ...data.encoded_plots]
         // Filter out the 'active' step and replace with 'complete'
         const filteredSteps = (lastMsg.steps || []).filter(s => s !== "Performing data analysis...");
@@ -134,15 +135,25 @@ export const ChatService = {
       // Status for intermediate nodes (completion)
       const statusMap: Record<string, string> = {
         "query_analyzer": "Query analysis complete.",
-        "planner": "Strategic plan generated.",
-        "coder": "Analysis code generated."
+        "planner": "Checklist generated.",
+        "delegate": "Task assigned.",
+        "reflector": "Result verified.",
+        "coder": "Analysis code generated.",
+        "executor": "Code execution complete.",
+        "searcher": "Web search complete.",
+        "summarizer": "Task summary generated."
       }
 
       if (name && statusMap[name]) {
         // Find and replace the active status if it exists
         const activeStatus = name === "query_analyzer" ? "Analyzing query..." : 
-                           name === "planner" ? "Generating strategic plan..." : 
-                           name === "coder" ? "Writing analysis code..." : null;
+                           name === "planner" ? "Generating high-level plan..." : 
+                           name === "delegate" ? "Routing task..." :
+                           name === "reflector" ? "Evaluating results..." :
+                           name === "coder" ? "Writing analysis code..." :
+                           name === "executor" ? "Executing code..." :
+                           name === "searcher" ? "Searching web..." :
+                           name === "summarizer" ? "Summarizing results..." : null;
         
         let filteredSteps = lastMsg.steps || [];
         if (activeStatus) {
@@ -159,9 +170,13 @@ export const ChatService = {
     if (type === "on_chain_start") {
       const startStatusMap: Record<string, string> = {
         "query_analyzer": "Analyzing query...",
-        "planner": "Generating strategic plan...",
+        "planner": "Generating high-level plan...",
+        "delegate": "Routing task...",
+        "reflector": "Evaluating results...",
         "coder": "Writing analysis code...",
-        "executor": "Performing data analysis..."
+        "executor": "Executing code...",
+        "searcher": "Searching web...",
+        "summarizer": "Summarizing results..."
       }
 
       if (name && startStatusMap[name]) {
@@ -170,9 +185,11 @@ export const ChatService = {
         const lastMsg = { ...newMessages[lastMsgIndex] }
 
         if (lastMsg && lastMsg.role === "assistant") {
-          // Avoid duplicate start messages
-          if (!(lastMsg.steps || []).includes(startStatusMap[name])) {
-            lastMsg.steps = [...(lastMsg.steps || []), startStatusMap[name]]
+          const currentSteps = lastMsg.steps || []
+          // Allow duplicate steps if it's a retry (cycle), 
+          // but avoid spamming the same active step multiple times in a row
+          if (currentSteps[currentSteps.length - 1] !== startStatusMap[name]) {
+            lastMsg.steps = [...currentSteps, startStatusMap[name]]
             newMessages[lastMsgIndex] = lastMsg
             return newMessages
           }