Description
autoskill is an agent skill that analyses a window of your own workflow history and proposes new skills from it. It reads from a local screenpipe daemon, which passively captures your activity, and it queries the daemon's local HTTP API for the time window you name, such as the last four hours or the last week. It clusters repeated research workflows, compares each pattern with the existing skills in the repository, and produces a staged folder of draft skills you can review, edit and promote. It is strictly on demand. It should only run when you ask for it, because it analyses sensitive local content, and it is not for one-off questions about screenpipe or for real-time screen queries. The skill requires a running screenpipe daemon. If the daemon is not running, it stops with install instructions. Its network use is limited: it talks to the local daemon, and to an LLM backend you configure. That can be a local LM Studio server by default, Anthropic's API if you opt in, or a bring-your-own-key gateway. Opt-in cloud backends receive redacted cluster summaries and matched skill descriptions. The listing shows it as free, and it has about 1,600 installs.…
Highlights
- Analyses a chosen window of local workflow history
- Clusters repeated research workflows
- Matches patterns against existing skills
- Stages draft skills for review
- Only runs when you ask; needs a local daemon
Ratings & Reviews
No ratings yet
Be the first to rate autoskill.
People also visit
autoskill FAQ
What does autoskill do?
It looks at a time window of your workflow history from a local screenpipe daemon, finds repeated research workflows, and drafts new skills from them.
What does it need to run?
A running screenpipe daemon on your machine. If the daemon is not running, the skill stops and gives install instructions.
Does it send my data anywhere?
The local daemon is queried on loopback. An LLM backend is used for analysis: LM Studio by default, or Anthropic or a bring-your-own-key gateway if you opt in. Opt-in cloud backends get redacted summaries.
When should I use it?
Only when you explicitly ask, for example to analyse your last day or week and propose skills. It is not for real-time screen queries.
How do I install it?
Run npx skills add https://github.com/k-dense-ai/scientific-agent-skills --skill autoskill.
Is it checked for security?
Its security audits show passes from Gen Agent Trust Hub and Socket, and a warning from Snyk.
What are the best autoskill alternatives?
Popular alternatives to autoskill include process-optimization, coding-agent and gh-issues. See all autoskill alternatives