langchain-ai / langchain-ai/langgraph

Invitation: review AgentGavel LangGraph adapter for governance benchmark provenance

Open
#8,992 0 comments 0 reactions 0 assignees View on GitHub
external
Dominant language
Python
Stars
41.8k
Forks
7.1k
Avg merge
23h 7m
Merged PRs (30d)
30

Description

## Context

[AgentGavel](https://github.com/agentgavel/agentgavel) is an open-source adversarial benchmark for AI agent **governance and security** (HITL gates, policy ceilings, provenance under attack) — not task-completion quality.

We ship an unofficial LangGraph-shaped sidecar adapter and would like LangGraph maintainers to review or contribute it so we can label scorecards `provenance=ratified` under our [ADR 007](https://github.com/agentgavel/agentgavel/blob/main/docs/adr/007-adapter-ratification.md) process.

## Ask

1. **Preferred:** Review or contribute [`adapters/langgraph/`](https://github.com/agentgavel/agentgavel/tree/main/adapters/langgraph) so Handshake / published scorecards can carry `provenance=ratified`.
2. **If you decline or do not engage:** After a 30-day public comment window we may grant `provenance=provisional` (distinct from ratified; expires in 180 days) based on an independent checklist. Community and maintainer comments belong on our tracking issue.

## Links

| | |
| --- | --- |
| Adapter | https://github.com/agentgavel/agentgavel/tree/main/adapters/langgraph |
| Tracking / comment window | https://github.com/agentgavel/agentgavel/issues/175 |
| Ratification policy | https://github.com/agentgavel/agentgavel/blob/main/docs/adr/007-adapter-ratification.md |
| Spec | https://github.com/agentgavel/agentgavel/blob/main/docs/RFC-0001.md |

## Honest dependency note

The in-tree adapter is a **minimal LangGraph-shaped stub** for harness CI (Oracle-backed chat completions + interrupt-style HITL). It does **not** currently depend on the `langgraph` PyPI package, so it does not exercise your full runtime. Ratification (or a maintainer-owned rewrite against real LangGraph APIs) is still the preferred path if you want scorecards to reflect the real product.

## Window

Public comment window for a possible `provisional` grant (if you do not ratify) **restarts from this outreach**: **2026-09-18 → 2026-10-18**. Details and objections: https://github.com/agentgavel/agentgavel/issues/175

Thank you for your time — happy to adjust the adapter or answer questions here or on #175.

— David Ndungu / AgentGavel (`@dndungu`)

Contributor guide

Open the contributing guide

Research direction

Start with the external adapters/langgraph/ tree, then read docs/adr/007-adapter-ratification.md and docs/RFC-0001.md; use issue #175 for the comment window and objections. Done means a maintainer review or contribution establishes ratified provenance, or the outcome is documented under the stated provisional process.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
ai, security
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Active
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.