F1-03
March 21, 2026 · View on GitHub
Objective
Turn the current integrations/langchain-python/ design scaffold into a functional Python integration built on top of the implemented Agent-DID Python SDK.
This checklist is the execution companion to:
docs/F1-03-LangChain-Python-Integration-Design.mddocs/F1-03-LangChain-Python-Technical-Plan.mdintegrations/langchain-python/README.md
Current Status
The blocking dependency has changed.
- Before: the integration was blocked by the absence of the Python SDK.
- Now: the Python SDK exists, has dedicated CI, and parity governance with the TypeScript SDK.
The remaining work is implementation work inside integrations/langchain-python/.
Implementation Phases
Phase 1 — Minimal Functional Integration
- Replace the placeholder
NotImplementedErrorfactory with a real integration factory. - Define the public integration config for Python.
- Implement current identity exposure.
- Implement DID resolution tool.
- Implement signature verification tool.
- Keep sensitive operations opt-in by default.
Phase 2 — Secure Optional Operations
- Implement HTTP signing as explicit opt-in.
- Implement payload signing as explicit opt-in.
- Keep private key material outside model-visible context.
- Keep key rotation disabled by default.
- Keep arbitrary signing disabled by default unless explicitly enabled.
Phase 3 — Context Injection and UX
- Inject DID, controller, active authentication method, and capabilities into agent context.
- Align naming and conceptual API with the JS integration where practical.
- Add a runnable example equivalent to the JS package quick start.
- Document LangChain Python version expectations.
Phase 4 — Validation
- Add Python tests for the integration package.
- Add tests for identity exposure.
- Add tests for DID resolution.
- Add tests for signature verification.
- Add tests for opt-in HTTP signing.
- Add regression coverage for secret isolation.
Phase 5 — Release Readiness
- Update
pyproject.tomlmetadata away fromdesign-scaffoldsemantics. - Update
README.mdfrom design scaffold language to implementation language. - Add package-level release checks aligned with
docs/SDK-Release-Checklist.md. - Add CI coverage if the package becomes executable and testable.
Phase 6 — Observability and Operationalization
- Add vendor-neutral observability hooks with secret redaction.
- Add structured JSON logging adapter.
- Add LangSmith adapter without changing the public factory.
- Support handler composition for fan-out to multiple sinks.
- Add runnable observability examples.
- Add a production-oriented opt-in example using environment configuration and a real model path.
- Document operational observability recipes in the package README.
Definition of Done
The LangChain Python integration is ready when all of the following are true:
create_agent_did_langchain_integration(...)is implemented.- The integration exposes identity, DID resolution, and verification tools.
- HTTP signing is opt-in and secret-safe.
- A runnable example exists.
- Automated tests exist.
- Documentation no longer describes the package as blocked by the Python SDK.
Current Closure Notes
- The package now includes opt-in key rotation.
- HTTP signing rejects insecure schemes and private or loopback targets by default.
- Dedicated CI coverage is expected in
.github/workflows/ci-langchain-python.yml. - The package now includes vendor-neutral observability, JSON logging, LangSmith adapters and handler fan-out.
- The package README now documents operational recipes for local debugging, structured logging, LangSmith and composed sinks.
- A production-oriented opt-in example now exists for real-model execution via environment variables.
Related Files
integrations/langchain-python/README.mdintegrations/langchain-python/pyproject.tomlintegrations/langchain-python/src/agent_did_langchain/__init__.pydocs/F1-03-LangChain-Python-Integration-Design.mddocs/F2-01-TS-Python-Parity-Matrix.mddocs/SDK-Release-Checklist.md