Detect repeated screen workflows and draft skill proposals from screenpipe captures.
npx skills add https://github.com/jasrajtulsi/GRAD-SCOPE --skill autoskill-jasrajtulsi
Please help me install this Agent Skill. Skill: autoskill Source: https://github.com/jasrajtulsi/GRAD-SCOPE/tree/main/.claude/skills/autoskill Command: npx skills add https://github.com/jasrajtulsi/GRAD-SCOPE --skill autoskill-jasrajtulsi
Autoskill turns your recent work patterns into proposed scientific-agent skills so you can spot repeated workflows, reduce manual reasoning, and grow a skill library from the way you actually work.
Ask autoskill to analyze a recent time window of your work and propose new or composable skills based on the repeated workflows it finds.
High-intent search queries and answers about installing and using this skill.
To turn repeated workflows into reusable skills, this tool analyzes recent screenpipe-captured activity, groups recurring app and window-title patterns, and drafts new skill proposals for uncovered sessions. It identifies repeated knowledge-work patterns and generates either a thin composition recipe or a full new SKILL.md draft.
Local embedding-based skill matching compares detected screen activity workflows against an existing skill library using local sentence-transformers. It classifies whether workflows represent reuse, composition, or novelty, keeping raw OCR data on the host machine without sending it externally.
Yes, screenpipe capture is required. The skill reads a recent time window of screenpipe-captured activity to detect repeated app combinations and window-title signatures, filtering out one-off sessions to identify workflows that warrant a drafted skill proposal.
Yes, privacy guardrails redact sensitive text before any data reaches the LLM backend. The workflow detection and embedding-based matching remain local to your machine, ensuring raw OCR data from screenpipe stays on the host while the LLM only processes redacted classifications.
To automate literature review skill drafting, the tool detects when you repeatedly move between Chrome, Zotero, and writing tools. It recognizes this window-title signature pattern and proposes an appropriate skill automatically, reducing the need to manually describe your workflow from scratch.
Limitations include the requirement for an active LLM backend to classify reuse, composition, or novelty, and dependence on screenpipe capture quality. One-off sessions are filtered out, meaning infrequent knowledge-work patterns will not generate a skill proposal.
Lord1EgyptDetect repeated research workflows from local screen activity and synthesize new skills.
dralkhDetect repeated research workflows from local screenpipe activity and draft skill proposals.
understudylabsIngest production LLM traces into redacted eval sets and cost profiles.
shushuznDiscover repeated research workflows from screen activity and draft new skill definitions.
