langchain - 💡(How to fix) Fix feat: Add AI Identity partner package for agent governance [1 comments, 2 participants]

Feature Request

Problem

LangChain agents making autonomous decisions (tool calls, API requests, financial actions) lack built-in governance infrastructure. There's no native way to:

• Enforce per-agent identity and scoped permissions at the gateway level
• Produce tamper-evident audit trails that prove what an agent did, when, and under which policy
• Verify after the fact that audit records haven't been altered (cryptographic chain integrity)

This becomes critical for teams deploying agents in regulated industries (finance, healthcare, legal) where "we have logs" isn't sufficient — you need forensic-grade evidence.

Proposed Solution

A langchain-ai-identity partner package at libs/partners/ai-identity/ that integrates AI Identity's governance layer into LangChain agents. Components:

• AIIdentityGovernanceMiddleware — Middleware for langchain_v1 agents with policy enforcement + HMAC audit
• AIIdentityChatOpenAI — Drop-in ChatOpenAI replacement that routes through the AI Identity gateway
• AIIdentityToolkit — Wraps tools with pre-execution policy checks
• AIIdentityCallbackHandler — Audit logging callback handler
• create_ai_identity_agent — Factory function to wire everything together

What's Already Built

• Full implementation with 30 unit tests (all passing)
• Published on PyPI as langchain-ai-identity 0.1.0
• PR #36590 was auto-closed due to missing issue link — ready to reopen once this issue is approved

Why a Partner Package

This follows the same pattern as other partner integrations (e.g., langchain-openai, langchain-anthropic). Agent governance is infrastructure that sits between the agent and upstream APIs — it's a natural integration point, not application logic.

Related PR

#36590 — ready to reopen with this issue linked