feat: add langgraph runtime

[sozercan]

Summary

• add runtime: langgraph registration, routing, fixtures, catalog entry, and docs
• add a LangChain/LangGraph adapter image that serves the existing AgentKit OpenAI /v1/chat/completions facade
• support stdio MCP tools through persistent langchain-mcp-adapters sessions with declared-env-only subprocess env
• add LangGraph conformance and guardrail tests, including MCP env isolation and token usage aggregation across tool-loop model calls

Validation

• go test ./... -race
• pytest -q in runtimes/langgraph (28 passed)
• pytest -q in runtimes/pydantic-ai (13 passed)
• pytest -q in runtimes/microsoft-agent-framework (22 passed)
• docker buildx build . -f runtimes/langgraph/Dockerfile -t agentkit-serve-langgraph:test --load
• make build-agentkit
• make build-test-agent RUNTIME=langgraph PLATFORM=linux/arm64
• smoke-tested /healthz and authenticated /v1/models for the built LangGraph image
• autoreview final run: clean, no accepted/actionable findings

[Copilot]

Pull request overview

Note

Copilot couldn't run its full agentic review because no GitHub Actions runner was available. Make sure your repository has a runner available to run Copilot's review, or add a copilot-setup-steps.yml file specifying one with the runs-on attribute. See the docs for more details.

This PR adds a new langgraph runtime adapter to AgentKit. It registers the runtime in the Go converter (const/spec/router), wires a LangChain/LangGraph Python adapter that serves the shared OpenAI /v1/chat/completions façade, supports stdio MCP tools via persistent langchain-mcp-adapters sessions with declared-env-only isolation, and adds fixtures, catalog entry, CI wiring, docs, and conformance/guardrail tests.

Changes:

• Register runtime: langgraph in the Go runtime registry, router, validator, and constants (with tests).
• Add the Python LangGraph adapter (agent_factory.py, __main__.py, Dockerfile, pyproject, tests) reusing the shared conformance suite and core façade.
• Update docs, Makefile, CI, fixtures, and catalog to build/test the LangGraph adapter alongside pydantic-ai and MAF.

Show a summary per file

File	Description
pkg/utils/const.go, runtime.go	Add RuntimeLangGraph constant and registry entry
pkg/utils/runtime_test.go, pkg/build/router_test.go, pkg/agentkit/config/config_test.go	Extend Go tests to cover the new runtime
runtimes/langgraph/agentkit_serve/agent_factory.py	Core adapter: model build, MCP tool loading, message/usage mapping, run loop
runtimes/langgraph/agentkit_serve/init.py, main.py	Package init and CLI binding
runtimes/langgraph/Dockerfile, pyproject.toml	Adapter image and dependency declarations
runtimes/langgraph/tests/*	Conformance re-export, conftest fixtures, guardrail tests
runtimes/common/agentkit_serve_common/conformance.py	Add LangChain/LangGraph SDK roots to the import-boundary check
runtimes/catalog/langgraph.yaml	Human-readable catalog entry
Makefile, .github/workflows/ci.yml	Build/test/smoke targets for the LangGraph runtime
README.md, runtimes/langgraph/README.md, test/foundry-hosted-agent/README.md	Documentation updates
test/agentkitfile-langgraph-hello.yaml, test/agentkitfile-langgraph-tools.yaml	New build fixtures

Review details

Tip

Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

• Files reviewed: 22/22 changed files
• Comments generated: 2
• Review effort level: Medium

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 52a34ff04f

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[Copilot]

Review details

• Files reviewed: 22/22 changed files
• Comments generated: 0 new
• Review effort level: Medium