feat(adapter): add LangGraph export and run adapter

docs/adapters/langgraph.md

@@ -0,0 +1,186 @@
+# LangGraph Adapter
+
+Complete mapping guide for converting between gitagent and LangGraph formats.
+
+## Overview
+
+[LangGraph](https://langchain-ai.github.io/langgraph/) is LangChain's graph-based agent orchestration framework for Python. It models agents as `StateGraph` objects where:
+
+- **Nodes** are Python functions that receive and return state
+- **Edges** are routing decisions (conditional or fixed)
+- **State** is a typed dictionary (`TypedDict`) carrying messages between nodes
+- **Tools** are Python functions decorated with `@tool`
+
+The gitagent LangGraph adapter enables:
+1. **Export**: Convert gitagent → LangGraph Python code (`agent.py`)
+2. **Run**: Execute gitagent agents using the generated Python agent
+
+## Installation
+
+```bash
+# LangGraph requires Python 3.11+
+pip install langgraph langchain-core
+
+# Provider-specific:
+pip install langchain-openai        # for GPT models
+pip install langchain-anthropic     # for Claude models
+pip install langchain-google-genai  # for Gemini models
+```
+
+## Field Mapping
+
+### Export: gitagent → LangGraph
+
+| gitagent | LangGraph | Notes |
+|----------|-----------|-------|
+| `SOUL.md` | `SYSTEM_PROMPT` (identity section) | Embedded as Python string constant |
+| `RULES.md` | `SYSTEM_PROMPT` (constraints section) | Appended to system prompt |
+| `DUTIES.md` | `SYSTEM_PROMPT` (SOD section) | Appended to system prompt |
+| `skills/*/SKILL.md` | `SYSTEM_PROMPT` (skills section) | Progressive disclosure, full instructions |
+| `tools/*.yaml` | `@tool` decorated functions | Stub implementations — fill in logic |
+| `knowledge/` (always_load) | `SYSTEM_PROMPT` (knowledge section) | Reference documents embedded |
+| `manifest.model.preferred` | LLM constructor (`ChatOpenAI`, `ChatAnthropic`, `ChatGoogleGenerativeAI`) | Auto-detected from model name prefix |
+| `compliance.supervision.human_in_the_loop: always` | `human_review_node` (blocks on every tool call) | Inserted between `agent` and `tools` nodes |
+| `compliance.supervision.human_in_the_loop: conditional` | `human_review_node` (blocks on risky tools) | Heuristic based on tool name keywords |
+| `compliance.supervision.human_in_the_loop: none` | No HITL node | Fully autonomous |
+| `manifest.version` | Docstring | Recorded for traceability |
+
+### Graph Structure
+
+**Without HITL:**
+```
+START → agent → (should_continue) → tools → agent (loop)
+                                  ↓
+                                 END
+```
+
+**With HITL (`always` or `conditional`):**
+```
+START → agent → human_review → (should_continue) → tools → agent (loop)
+                                                  ↓
+                                                 END
+```
+
+## Model Resolution
+
+| gitagent `model.preferred` | LangGraph import | Class |
+|----------------------------|------------------|-------|
+| `claude-*` or `anthropic/*` | `langchain_anthropic` | `ChatAnthropic` |
+| `gpt-*`, `o1*`, `o3*`, `openai/*` | `langchain_openai` | `ChatOpenAI` |
+| `gemini-*` or `google/*` | `langchain_google_genai` | `ChatGoogleGenerativeAI` |
+| *(other)* | `langchain_openai` | `ChatOpenAI` |
+
+## Usage Examples
+
+### Export to LangGraph
+
+```bash
+# Export to stdout
+gitagent export --format langgraph -d ./my-agent
+
+# Save to file
+gitagent export --format langgraph -d ./my-agent -o agent.py
+```
+
+**Output Structure:**
+```
+# === agent.py ===
+"""LangGraph agent generated from gitagent manifest…"""
+from langchain_openai import ChatOpenAI
+from langgraph.graph import StateGraph, END
+…
+
+# === requirements.txt ===
+langgraph>=0.2.0
+langchain-core>=0.3.0
+langchain-openai>=0.3.0
+python-dotenv>=1.0.0
+
+# === .env.example ===
+OPENAI_API_KEY=your-openai-api-key
+```
+
+### Run with LangGraph
+
+```bash
+# Interactive mode
+gitagent run --adapter langgraph -d ./my-agent
+
+# Single-shot mode
+gitagent run --adapter langgraph -d ./my-agent --prompt "Summarise the latest SEC filings"
+```
+
+The runner:
+1. Generates `agent.py`, `requirements.txt`, and `.env.example` in a temp workspace
+2. Creates a Python virtual environment
+3. Installs dependencies via `pip install -r requirements.txt`
+4. Executes `python agent.py [--prompt "…"]`
+5. Cleans up the temp workspace on exit
+
+### Running the generated agent directly
+
+```bash
+# After export:
+cp .env.example .env && vim .env   # add your API key
+pip install -r requirements.txt
+python agent.py                     # interactive REPL
+python agent.py --prompt "Hello"    # single-shot
+```
+
+## Compliance Mapping
+
+| gitagent `human_in_the_loop` | LangGraph behaviour |
+|------------------------------|---------------------|
+| `always` | `human_review_node` inserted — user must type `y` to approve **every** tool call |
+| `conditional` | `human_review_node` inserted — approval required only for write/delete/send/post operations |
+| `advisory` | No HITL node — advisory note added in system prompt only |
+| `none` | No HITL node — fully autonomous |
+
+## Implementing Tools
+
+The adapter generates stub `@tool` functions from `tools/*.yaml`. You need to fill in the implementation:
+
+```python
+# Before (generated stub)
+@tool
+def search_regulations(query: str) -> str:
+    """Search regulatory database."""
+    raise NotImplementedError("Implement search_regulations tool")
+
+# After (your implementation)
+@tool
+def search_regulations(query: str) -> str:
+    """Search regulatory database."""
+    results = my_db.search(query)
+    return "\n".join(r.text for r in results)
+```
+
+## Generated File Reference
+
+### `agent.py`
+
+| Section | Description |
+|---------|-------------|
+| `SYSTEM_PROMPT` | Full agent identity + rules + skills as string constant |
+| `TOOLS` | List of `@tool`-decorated functions (stubs from `tools/*.yaml`) |
+| `AgentState` | `TypedDict` with `messages: Annotated[list[BaseMessage], add_messages]` |
+| `agent_node` | Calls LLM with tools bound; prepends system prompt |
+| `should_continue` | Routes to `"tools"` if last message has tool calls, else `END` |
+| `human_review_node` | (HITL only) Prompts user for approval before tool execution |
+| `build_graph` | Wires nodes and edges into a compiled `StateGraph` |
+| `main` | CLI entry point: `--prompt` for single-shot, interactive REPL otherwise |
+
+### `requirements.txt`
+
+Minimal set of pip packages needed to run the agent.
+
+### `.env.example`
+
+Template for environment variables (API keys). Copy to `.env` and fill in.
+
+## Notes
+
+- Tools are generated as **stubs** — the adapter cannot infer implementation from YAML declarations alone. Fill in the function body before use.
+- The runner requires **Python 3.11+** and **pip** on PATH.
+- For production use, replace the `tmpdir`-based approach with a persistent project directory.
+- LangGraph does not natively support gitagent's agent versioning or branch deployment patterns — those remain in the git layer.

src/adapters/index.ts

@@ -9,3 +9,4 @@ export { exportToOpenCodeString, exportToOpenCode } from './opencode.js';
 export { exportToCursorString, exportToCursor } from './cursor.js';
 export { exportToGeminiString, exportToGemini } from './gemini.js';
 export { exportToCodexString, exportToCodex } from './codex.js';
+export { exportToLangGraphString, exportToLangGraph } from './langgraph.js';