# tigerless-labs/autoharness

Autoharness — a self-learning skill layer for Claude Code — distills skills from your real sessions, updates them as you work, and prunes the ones that stop getting used. No daemon, no benchmark.

Repository: https://github.com/tigerless-labs/autoharness
Canonical: https://ross.abutalabs.com/products/autoharness
Homepage: https://github.com/tigerless-labs/autoharness
Language: Python
License: MIT
License Family: permissive
Topics: agent-skills, claude-code, claude-code-plugin, llm-agents, python
Last push: 2026-07-28T21:33:18+00:00

## Health v2 (maintenance only)
Score: 56/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 94, release rhythm 35, longevity 6
- inputs: {"age_days": 85, "days_push": 36, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, young
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1263, forks 80 (observed 2026-08-28T04:04:10.261444+00:00)

## What it is
Autoharness is a self-learning skill layer for Claude Code that distills reusable skills from your real coding sessions, merges near-duplicates, updates skills in use, and prunes unused ones. It runs entirely as a zero-dependency Python plugin with hooks and an MCP server, touching only the skills it generated itself.

## Use cases
- automatically learn skills from my Claude Code sessions
- keep my agent skill library clean without manual curation
- distill repeated workflows into reusable Claude Code skills
- merge duplicate agent skills automatically
- prune agent skills that stop being used
- improve coding agent performance with a self-maintaining harness
- capture session learnings into .claude/skills

## When to choose
- you use Claude Code regularly and want skills to accumulate and maintain themselves from real work
- you want skill consolidation and pruning without running a separate daemon or eval benchmark
- you want a zero-dependency Python plugin that only manages skills it created itself

## When to avoid
- you don't use Claude Code or a compatible agent harness
- you need skills that work across multiple different AI coding tools
- you prefer fully hand-written, manually curated skill files
- you're on Windows, which isn't listed as a supported platform

## Facets
- artifact type: plugin
- maturity: active
- function: agent-framework, prompt-engineering, developer-tools, workflow-automation
- domain: developer-tools, large-language-models
- platform: python, cli
- tags: claude-code, claude-code-plugin, agent-skills, self-learning, skill-management, llm-agents, mcp, ai-agents, automation, linux, macos

## Member repositories
- tigerless-labs/autoharness (main) score 56

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:10.261444+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T05:04:11.782545+00:00, confidence not recorded.
  - readme: https://github.com/tigerless-labs/autoharness (fetched 2026-08-28T04:04:10.261444+00:00, sha 817b88156cf5)
  - homepage: https://github.com/tigerless-labs/autoharness (fetched 2026-08-29T12:16:38.997728+00:00, sha 50ee9a2478d0)
- Data as of 2026-08-30T08:39:29.467469+00:00.