Entelligence's agentplay replays your coding agent sessions as a map and a timeline

Entelligence · Oct 14, 2026

Entelligence's agentplay is an open-source (Apache-2.0) session replay for coding agents. It runs locally and, per Entelligence, session data stays on your machine; the only network call is an optional email signup. Made public on Oct 14, it reads the session logs that Claude Code, Codex, Cursor, GitHub Copilot, OpenCode and Kiro already keep on your computer and turns each session into a map of the files, commands, MCP tools, models and subagents it touched, plus a timeline that marks retries, slow calls, errors, context compactions, model switches and opened PRs. It needs Python 3.11+ and no account or API key.

How you’d use it: Replay a coding agent's session to see which commands it retried and why.

Sources

• Original source: Entelligence ↗
• Entelligence's agentplay on GitHub (Apache-2.0) ↗

← All AI news