Ai Skillweave

ai-skillweave

Cross-harness skill delivery: **2,652 unique Agent Skills** from 14 upstream repos, synced into Claude/Codex/OpenClaw/Pi/Copilot/Hermes. Auto-learns from your corrections via hooks and a 4-stage batched distillation pipeline.

One-command setup for all your Ollama agent harnesses: proper MCP servers, web tools, harness-specific configs — all portable and reproducible. Learns from your corrections automatically via real-time hooks and a batched learning pipeline that distills conversations into skills 30-50× faster than naive LLM calls.

**2,652 unique skills, delivered to all 6 harnesses** from 14 upstream repos: [OpenClaw-Medical](https://github.com/FreedomIntelligence/OpenClaw-Medical-Skills) (896 clinical), [operon](https://github.com/swaruplab/operon) (556 protocols), [bioSkills](https://github.com/GPTomics/bioSkills) (546, 63 categories), [ECC](https://github.com/affaan-m/everything-claude-code) (272), [SciAgent](https://github.com/jaechang-hits/SciAgent-Skills) (197), [ToolUniverse](https://github.com/mims-harvard/ToolUniverse) (150 drug discovery), [K-Dense](https://github.com/K-Dense-AI/scientific-agent-skills) (147), [ClawBio](https://github.com/ClawBio/ClawBio) (90), [DeepMind science-skills](https://github.com/google-deepmind/science-skills) (37), [Bipartite](https://github.com/matsen/bipartite) (37), [BioNeMo](https://github.com/NVIDIA-BioNeMo/bionemo-agent-toolkit) (35 GPU-accelerated bio), [Nature-Paper](https://github.com/Boom5426/Nature-Paper-Skills) (18 manuscript), [Anthropic](https://github.com/anthropics/skills) (17), [life-sciences](https://github.com/anthropics/life-sciences) (6). See [`docs/SKILLS-CATALOG.md`](docs/SKILLS-CATALOG.md) for the full per-source breakdown.

---

Skill Sources

Each source follows the [Agent Skills](https://agentskills.io/) standard (SKILL.md with `name` + `description` frontmatter):

Source	On-disk	Net-new	Notes
[OpenClaw-Medical](https://gi