77k Stars for a Repo That Just Stops Agents Taking the Shortcut
Addy Osmani's agent-skills isn't a model or a framework — it's 24 workflows that force a coding agent to write the spec, the test, and the security review it would otherwise skip. The star count is the market pricing in what actually separates a demo from production.
NeuroX AI · July 13, 2026
Addy Osmani's agent-skills crossed 77.7k GitHub stars — up from about 31k in early May. It ships no model and no framework. It's 24 skills: plain workflows that make a coding agent do the work it would otherwise skip.
The repo's own diagnosis is the whole reason it went viral: "AI coding agents default to the shortest path — which often means skipping specs, tests, security reviews, and the practices that make software reliable." Left alone, an agent optimizes for the answer that looks done, not the one that survives production. agent-skills encodes the judgment a senior engineer applies without thinking — spec-driven development, test-driven development, code review, security hardening, observability — into 23 lifecycle steps the agent is forced to walk through, plus one meta-skill for using the rest.
Notice what the market is starring here. Not a smarter model — the models are already good enough. It's the discipline layer around them: define, plan, build, verify, review, ship. The same seams that decide whether a prototype survives contact with real systems.
That's the entire NeuroX thesis in someone else's README. The moat was never the model. It's the process you wrap around it — and now 77,000 developers agree.