tigerless-labs/autoharness
- Source
- GitHub
- First trending
- Category
- Skills
- GitHub stars
- 5,316
- Main language
- Python
This page introduces an external open-source repository. It is not an HDATF product.
What it does
A self-learning skill layer for Claude Code. It distills skills from real sessions, merges same-scenario ones, updates them in use and prunes unused ones, touching only the skills it wrote.
How it helps ATF
When considering how Harness could maintain agent skills, its check of whether later turns actually follow a skill instead of a benchmark score, and its record of which skill absorbed which, are worth comparing.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
More in this category
- steipete/agent-scripts
Shared agent instructions, reusable workflow skills, portable helper scripts, and local validation hooks for multiple development workspaces. - newliver666/apk-reverse
An Agent Skill for Android APK reverse engineering, covering debloating, ad removal, dex patching, repacking and runtime or server analysis. It pairs a short SKILL.md with parameterized scripts and references that load only when a step needs them. - coji/natural-japanese
An agent skill for writing and revising readable Japanese work documents. It combines writing rules with morphological-analysis-based linting to identify formulaic expressions and monotonous prose. - chubbyguan/chubbyskills
Agent skills and command-line tools that save videos, podcasts, articles, and local documents as Markdown in a searchable local knowledge base. It supports source-linked material exports and an optional MCP service. - ai-evals-course/evals-skills
A set of skills that guide AI coding agents through building product-specific evals, with an entry point skill that routes either to auditing an existing eval pipeline or to finding errors in unanalyzed traces.
Only repositories in the ranked Trendshift lists are included, and the lists are used only to find candidates. We do not copy their ranks. Descriptions, licenses and star counts come from each GitHub repository. The notes are our own reading. We have not tested these projects, and a place on a trending list does not prove quality.