Autoskill

This one watches your screenpipe history, clusters the workflows you actually repeat, and drafts new Claude skills for the patterns that aren't already covered by the 135 skills in the repo. You run it on demand with a time window, it analyzes locally captured screen data, compares clusters against existing skills using local embeddings, then hands an LLM (defaults to LM Studio with Gemma) the redacted summaries to decide whether to compose existing skills or write new ones. Proposals land in a staging folder you review before promoting. The privacy model is solid: raw OCR never leaves your machine, screenpipe filters sensitive apps at capture time, and the detection pipeline is fully local. Useful if you want your skill library to grow from actual usage rather than guessing what you might need.

skills.sh installs: 1.9K
GitHub stars: 56.6K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add k-dense-ai/scientific-agent-skills --skill autoskill --agent claude-code

Installs into .claude/skills of the current project.

Files

autoskill

Requires a running screenpipe daemon. This skill has no alternate data source — it reads exclusively from the local screenpipe HTTP API (default http://localhost:3030). If the daemon isn't running, run() raises ScreenpipeUnreachable with install instructions.

Network access & environment variables. This skill makes authenticated HTTP requests to (a) the user's local screenpipe daemon on loopback, and (b) the user-configured LLM backend — one of http://localhost:1234/v1 (LM Studio, default), https://api.anthropic.com (opt-in Claude), or a user-supplied BYOK Foundry gateway. The adapters read SCREENPIPE_TOKEN, LM_API_TOKEN, ANTHROPIC_API_KEY, and FOUNDRY_API_KEY for the corresponding configured service. HTTPX also honors its standard proxy and CA environment settings; sentence-transformers uses Hugging Face cache/download settings. Opt-in cloud backends receive redacted cluster summaries and matched skill descriptions. Embedding-model installation may also download public model files; local inference does not imply zero network access.