autoharness
Building the self-learning layer next-gen agents will ship with, learning skills from real work.
11164stars

An open AI lab researching the architecture of next-generation agents.
Tools that change how your coding agent works — and show you what it is actually doing.
Building the self-learning layer next-gen agents will ship with, learning skills from real work.
11164stars

Usage logs tell you what a turn cost; Cost X-ray tells you why, attributing every token in the context to its source.
5797stars

Long-term memory that gives any agent a retrieval engine's ranking over a filesystem it can just read.
4323stars

Skills that put the agent to work on everything around the code — design, research, distribution.
Turns your PRs and the day's hot threads into topics, X and Reddit post drafts, and reply drafts — drafts only, never posts for you.
780stars

A crawler's-eye audit of your site's SEO and GEO foundation: 30 deterministic checks, pass or fail.
992stars

Feed your agent papers and half-formed ideas; it links them into a system design you can defend. An Agent Skill for Claude Code and any SKILL.md-compatible agent.
236stars

Finds the AI papers 28 labs put on arXiv over any date range, and separates the ones a company led from the ones it only appears on.
226stars

Free, open-source influencer discovery: finds creators to partner with across 15 major platforms — TikTok, Instagram, X and more — with their public contact info.
226stars

Tools for safer and more reliable AI development workflows.
PHI boundary reports for healthcare and insurance AI workflows — across prompts, RAG, tools, memory, and logs.
204stars

Lightweight, rule-based checks for missing, weak, and uncovered tests in pull requests.
193stars

More in the works — built in the open, shipped fast.