Small executable "skills" — glue code that connects LLM-based agents to tools and external data — are where most innovation and most risk live in agent ecosystems. This repository preserves every archived version of those skills from ClawHub as a forensic snapshot (repository created on 2026-01-06 and widely starred), which makes it useful for studying evolution, provenance, and threat patterns across an agent marketplace.
What Sets It Apart
- Complete-version archive: stores historical versions rather than only the latest release — so you can trace when behavior or interfaces changed, which helps reproducibility and regression analysis. (So what: you can reconstruct an agent's exact runtime environment from a given date.)
- Research & security value: retains items flagged as suspicious or malicious for short-term analysis rather than immediate removal. (So what: enables malware analysis, supply-chain forensics, and detection-rule development.)
- Lightweight cataloging for agents: organized as an archival dump tied to ClawHub/ClawDBot usage patterns rather than a curated plugin marketplace. (So what: it's better for bulk research than for safe end-user consumption.)
- Public, easily referenced snapshot: GitHub-hosted and discoverable (not a gated dataset). (So what: straightforward to cite or include in dataset builds, but be mindful of licensing and safety.)
Who It's For & Tradeoffs
Great fit if you are a security researcher, academic studying agent ecosystems, or developer building tooling for vetting/curation — you get versioned artifacts and provenance signals useful for audits and model-tooling research. Look elsewhere if you want a curated, safety-reviewed skill store for production use — this repo intentionally preserves risky items and explicitly warns against running unvetted code. Practical cautions: never execute skills from the archive without sandboxing and manual review; treat the repository as a historical dataset, not a drop-in plugin source.
Where It Fits
This is an archival complement to curated marketplaces and verified plugin stores: use it when your goal is historical analysis, threat hunting, or building vetting pipelines. For day-to-day agent extensions, prefer official, reviewed distribution channels and the live ClawHub site for safer downloads.
