Logslim, an open-source tool, compacts verbose test and build logs by 80–95% to help both developers and AI agents read failures faster.

What happened

Logslim strips ANSI colors, deduplicates repeated warnings, collapses stack frames from node_modules, and extracts structured errors (file, line, message) from test and build output. It runs as a CLI wrapper, GitHub Action, or MCP server for Claude Code and Cursor, and is available on MIT license with no account or API key required.

Why it matters

When CI fails or an AI agent runs tests, reading 500–3000 lines of noise (progress bars, duplicate warnings, vendor stack traces) wastes time and tokens. Logslim reduces that to 25–55 lines while preserving the actual failure and fix hints, letting humans spot the problem immediately and agents focus compute on diagnosis rather than noise.

What to watch

Measured token savings range from 64% (Pytest with 25 identical failures) to 92% (Jest with warning spam and one failure). Every removed section is marked with a note so the agent or developer can re-run the raw command if full logs are needed; the tool also attaches fix cards for known error codes (TypeScript TS2339, npm ERESOLVE, Node ECONNREFUSED) with 2–4 short fix steps per code.