Highlights
- Separate structural validation from semantic Skill quality assessment.
- Generate human-readable and machine-readable quality annotations for every compiled bundle.
- Block generic, repetitive, low-evidence, or underspecified Skills before installation.
- Bind installation approval to the reviewed bundle contents and destination.
- Expose 23 dependency-free MCP tools with 24 passing behavioral tests.
Try it
git clone https://github.com/sdfdu/evopilot.git
cd evopilot
python plugins/evopilot/scripts/evopilot.py doctor
python plugins/evopilot/scripts/evopilot.py demo --destination ./evopilot-demoSee RELEASE_NOTES.md for the complete history.