Skill-harness is a Python evaluation tool for reusable AI agent skills. It assesses whether a skill earns its place by returning clear KEEP, CUT, or CAN'T-TELL-YET verdicts instead of manufactured scores. It offers first-class support for Claude Code and is designed for developers building robust agent systems.
skill-harness sits in PulseGate's AI & ML category. It focuses on determining whether a reusable AI agent skill is valuable enough to keep in a library without relying on arbitrary numerical scores. skill-harness is an open-source project aimed at AI developers. The project is open source (MIT). It runs on the command line.
MrBinnacle builds and maintains skill-harness, and the product first shipped in 2026. The project is developed in the open on GitHub with 196 commits in the last 90 days. Among its 3 catalogued features are skill evaluation, binary decisions, and Claude Code support.
Latest indexed changes and source events
skill-harness verified by the PulseGate indexer
Other apps tracked under the same category.
