{"_id":"@artale/pi-gepa","_rev":"3-531efe05c6aa993bd33a2917e14ee399","name":"@artale/pi-gepa","dist-tags":{"latest":"1.2.0"},"versions":{"1.0.0":{"name":"@artale/pi-gepa","version":"1.0.0","keywords":["pi-package"],"author":{"name":"artale"},"license":"MIT","_id":"@artale/pi-gepa@1.0.0","maintainers":[{"name":"artale","email":"arosstale1@gmail.com"}],"homepage":"https://github.com/arosstale/pi-gepa#readme","bugs":{"url":"https://github.com/arosstale/pi-gepa/issues"},"pi":{"extensions":["src/extension.ts"]},"dist":{"shasum":"4911588a9fd816f41c70a62b75b4dd9494d7ea85","tarball":"https://registry.npmjs.org/@artale/pi-gepa/-/pi-gepa-1.0.0.tgz","fileCount":3,"integrity":"sha512-fbgOjnbwmP0rtiAN1I1/AMa6+0pE8G+zkjGtn4eRVCIv1wyX/KYUtnUqMHdwzbaVTJRe1MSpWEgUCMBjZcwwug==","signatures":[{"sig":"MEYCIQDqMUnNTDp8zleIGJ9o4U/skqbagCTYdiXDqSr4RcClPQIhAM/gc9K8zYAnTJlb41E12IaTFEH/n9UGgnLVblwlyd8n","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"unpackedSize":8762},"gitHead":"3a68a5186879bbf5b8ea1bb0a09338ddf4197840","_npmUser":{"name":"artale","email":"arosstale1@gmail.com"},"repository":{"url":"git+https://github.com/arosstale/pi-gepa.git","type":"git"},"_npmVersion":"10.9.4","description":"Genetic Evolution for Prompts and Agents (GEPA). Evolve Pi skills and prompts through mutation, crossover, and selection.","directories":{},"_nodeVersion":"22.22.0","_hasShrinkwrap":false,"peerDependencies":{"@mariozechner/pi-coding-agent":"*"},"_npmOperationalInternal":{"tmp":"tmp/pi-gepa_1.0.0_1773496542968_0.02672264415940262","host":"s3://npm-registry-packages-npm-production"}},"1.1.0":{"name":"@artale/pi-gepa","version":"1.1.0","keywords":["pi-package"],"author":{"name":"artale"},"license":"MIT","_id":"@artale/pi-gepa@1.1.0","maintainers":[{"name":"artale","email":"arosstale1@gmail.com"}],"homepage":"https://github.com/arosstale/pi-gepa#readme","bugs":{"url":"https://github.com/arosstale/pi-gepa/issues"},"pi":{"extensions":["src/extension.ts"]},"dist":{"shasum":"6715d0e3f971b83303d6240f18de395e3e49ed56","tarball":"https://registry.npmjs.org/@artale/pi-gepa/-/pi-gepa-1.1.0.tgz","fileCount":3,"integrity":"sha512-DTZ8z/e5aIKeulm1wv5+4qx78W/o7HFMHEn49UiyCgvw7EhM7wWbxWeZ/d04TBSuo3JtsBkFIyIoe31BAceT2A==","signatures":[{"sig":"MEYCIQChyEyBw/V7l9KeyqOiVNwfjoUguGoErl8gtW03RihnRQIhAO/vctLCh4JTw9euhecVko5wOxBmDoZrchSb4LmllYLK","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"unpackedSize":8770},"gitHead":"aae9d70707a8029ae2672dc63819b50612d5150b","_npmUser":{"name":"artale","email":"arosstale1@gmail.com"},"repository":{"url":"git+https://github.com/arosstale/pi-gepa.git","type":"git"},"_npmVersion":"10.9.4","description":"Genetic Evolution for Prompts and Agents (GEPA). Evolve Pi skills and prompts through mutation, crossover, and selection.","directories":{},"_nodeVersion":"22.22.0","_hasShrinkwrap":false,"peerDependencies":{"@mariozechner/pi-coding-agent":"*"},"_npmOperationalInternal":{"tmp":"tmp/pi-gepa_1.1.0_1773499076762_0.7731150263839739","host":"s3://npm-registry-packages-npm-production"}},"1.2.0":{"name":"@artale/pi-gepa","version":"1.2.0","description":"Skill scanner and quality scorer for Pi. Scans, scores, and identifies weak skills. Pairs with pi-evolve for mutation.","keywords":["pi-package"],"license":"MIT","author":{"name":"artale"},"repository":{"type":"git","url":"git+https://github.com/arosstale/pi-gepa.git"},"pi":{"extensions":["src/extension.ts"]},"peerDependencies":{"@mariozechner/pi-coding-agent":"*"},"_id":"@artale/pi-gepa@1.2.0","gitHead":"84973487a05740ed46f6bd4a90a84d6bbe219dbc","bugs":{"url":"https://github.com/arosstale/pi-gepa/issues"},"homepage":"https://github.com/arosstale/pi-gepa#readme","_nodeVersion":"22.22.0","_npmVersion":"10.9.4","dist":{"integrity":"sha512-MkFHT5MS4BUM00f8oVQlpaDVV3Etk1tw4E77NpFHwyQDiJSyRn2fz5HkoKbZRcSIV52PH0qyjFlrLpx+5nB0CA==","shasum":"7f7576395eff39e2f9fe2b88e0f08f6ee84535a2","tarball":"https://registry.npmjs.org/@artale/pi-gepa/-/pi-gepa-1.2.0.tgz","fileCount":3,"unpackedSize":9218,"signatures":[{"keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U","sig":"MEUCIAwp1sBegjMWzVCGDKFTiuca11roIQdYlf/uDlgLsK/aAiEAqTUEFfeYcrZ35vd7PCnDwggWTIGl+ysp5iqtL8+Zj9c="}]},"_npmUser":{"name":"artale","email":"arosstale1@gmail.com"},"directories":{},"maintainers":[{"name":"artale","email":"arosstale1@gmail.com"}],"_npmOperationalInternal":{"host":"s3://npm-registry-packages-npm-production","tmp":"tmp/pi-gepa_1.2.0_1773504227675_0.41899824423264875"},"_hasShrinkwrap":false}},"time":{"created":"2026-03-14T13:55:42.846Z","modified":"2026-03-14T16:03:47.942Z","1.0.0":"2026-03-14T13:55:43.107Z","1.1.0":"2026-03-14T14:37:56.907Z","1.2.0":"2026-03-14T16:03:47.831Z"},"bugs":{"url":"https://github.com/arosstale/pi-gepa/issues"},"author":{"name":"artale"},"license":"MIT","homepage":"https://github.com/arosstale/pi-gepa#readme","keywords":["pi-package"],"repository":{"type":"git","url":"git+https://github.com/arosstale/pi-gepa.git"},"description":"Skill scanner and quality scorer for Pi. Scans, scores, and identifies weak skills. Pairs with pi-evolve for mutation.","maintainers":[{"name":"artale","email":"arosstale1@gmail.com"}],"readme":"# pi-gepa\n\nGenetic Evolution for Prompts and Agents. Evolve Pi skills and prompts through mutation, crossover, and selection.\n\n## Install\n\n```bash\npi install npm:@artale/pi-gepa\n```\n\n## What it does\n\nGEPA applies evolutionary algorithms to prompt/skill optimization:\n\n1. **Mutate** — create variants of a skill with targeted changes\n2. **Crossover** — combine strengths of two skills\n3. **Select** — evaluate variants against a fitness function\n4. **Evolve** — repeat until convergence\n\n## Results\n\nTested on real Pi skills:\n- `canvas-design`: **−69% token reduction** while maintaining quality\n- `pptx`: **−60% token reduction**\n- `xlsx`: **−63% token reduction**\n\n## Commands\n\n```\n/gepa evolve <skill>         — start evolution of a skill\n/gepa status                 — show evolution state\n/gepa history                — past evolution results\n```\n\n## How it works\n\nUses the LLM itself as the mutation operator. Each generation:\n1. Reads the current skill SKILL.md\n2. Generates N mutated variants (shorter, restructured, different examples)\n3. Evaluates each variant on a benchmark task\n4. Selects the best performer\n5. Repeats for M generations\n\nThe fitness function measures: output quality + token efficiency + instruction clarity.\n\n## License\n\nMIT\n","readmeFilename":"README.md"}