Custom Agent Integration
Integrate custom or internal coding agents with Git AI telemetry collection.
Many engineering teams build their own coding agents — internal tools tailored to proprietary codebases, domain-specific workflows, or custom orchestration layers. Git AI tracks code from any agent, not just the ones with built-in support.
Add Your Agent — Full Integration Guide
How It Works
Git AI uses the git-ai checkpoint command to mark file changes as
AI-authored. Any agent that calls checkpoint before and after editing
a file gets full telemetry support: AI authorship tracking, prompt and
session traces, model attribution, and lifecycle analytics through the
Git AI for Teams dashboard.
There are two integration paths:
- Direct integration — the agent calls
git-ai checkpointdirectly from its agent loop. Best for getting started quickly. - Hooks and custom presets — the agent exposes a hook system (like
BeforeEdit/AfterEdit) and contributes a custom preset to Git AI. Best for agents that already have a hook architecture.
Both paths use the same underlying checkpoint mechanism and produce
identical telemetry.
Getting Help
The core maintainers are available to help integrate custom agents. Schedule a call to get hands-on support.