What is agentlogs?

AgentLogs is a self-hostable platform that gives teams visibility into AI coding agent sessions. It captures transcripts from several popular agents, automatically linking them to git commits so teams can see which session produced which code. The web dashboard shows activity metrics, agent and model usage, and per-member breakdowns. Sessions can be shared with the team for learning effective prompts and workflows. All transcripts are scanned for secrets (1,600+ patterns) and redacted locally before upload. It runs as a Docker container or standalone binary using SQLite for storage and BetterAuth for GitHub/GitLab OAuth authentication. The project is licensed under FSL-1.1-Apache-2.0 (source-available, converting to Apache-2.0 after two years).

Should you self-host agentlogs?

agentlogs scores 42/100 on selfhost.directory (how scoring works) and ranks #122 of 166 in Developer Tools. It has 212 GitHub stars.

It shipped 13 stable releases in the past year. The latest, server-v0.1.4, came out on 10 September 2026. New versions land roughly every 1 days.

Setup is moderate: expect some configuration beyond a single command. An official Docker image is available. It is Other-licensed and written mostly in Typescript.

Its closest alternative on selfhost.directory is openhands (87/100, 92 094 stars). It scores higher overall, so compare the two before you pick. See agentlogs vs openhands side by side.

Features

AI coding agent transcript capture
AI-generated transcript summaries
CLI Tool
Git commit linking
GitHub OAuth authentication
GitLab OAuth/OIDC authentication
Multi-agent plugin support
Secret redaction
Session Sharing
Team dashboard
Visibility Controls

FAQ

How does AgentLogs compare to git-ai or agent-trace?

Tools like git-ai and agent-trace focus on tracking AI attribution at the code level (which lines were AI-generated). AgentLogs focuses on the session and prompt level, capturing full context of how code was created. They are seen as complementary with plans to integrate.

Which coding agents are supported?

AgentLogs supports Claude Code, Codex CLI, OpenCode, Pi, and Cline. Each has its own lightweight plugin, with more agents being added actively.

Who can see my transcripts?

You control visibility per transcript: Private (only you), Team (you and team members), or Public (anyone with the link). Defaults: public repos are public, private repos are team-visible, sessions without a repo are private.

How does AgentLogs protect my secrets and sensitive data?

AgentLogs automatically scans all transcripts for secrets before uploading using 1,600+ detection patterns covering API keys, tokens, passwords, and database credentials. Detected secrets are redacted entirely on your machine and never leave in plain text.

How does AgentLogs capture my sessions?

AgentLogs uses lightweight plugins for each coding agent. Install a plugin with a single command and it captures transcripts at the end of each session, automatically linking them to git commits.

Is AgentLogs source-available?

Yes, the entire codebase (CLI, web app, plugins) is on GitHub under FSL-1.1-Apache-2.0 and can be self-hosted with the official Docker image or standalone binaries.