---
name: ager-to-langchain-deep-agents
description: Translate a validated AGER/OKF AgentGraph into a LangChain Deep Agents project (skills from AgentNodes, FilesystemBackend).
---

# AGER → LangChain Deep Agents

This plugin **compiles**. It does not author AGER. Authoring is `okf-agent-graph`.

## Mapping

| AGER | LangChain Deep Agents |
| --- | --- |
| OrchestratorAgent | create_deep_agent system / lead |
| Worker / Judge / Synthesizer | skills/<id>/SKILL.md |
| LoopPolicy | documented stop conditions in ager-run skill |
| ScratchPad | FilesystemBackend virtual FS |
| Tool | skill tool sections |

## Steps

1. Locate the AGER bundle (`--bundle`, `sample-ager`, or cwd).
2. If `okf-agent-graph` is installed, run `ager-validate` first. If it fails, stop.
3. Run the deterministic emitter:

```bash
python3 scripts/emit.py --bundle <AGER_ROOT> --out <OUT>
```

4. Report written paths. Expected: pyproject.toml, agent.py, skills/*/SKILL.md, README.md.
5. Never claim production-ready without tests.

## References

- `references/mapping.md`