k-dense-ai/scientific-agent-skills/skill-guides

Claude Code and agent skill by k-dense-ai/scientific-agent-skills

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Install

npx vskill@latest install k-dense-ai/scientific-agent-skills/skill-guides

SKILL.md

The same SKILL.md was published earlier as k-dense-ai/scientific-agent-skills/skill-guides.

All skill guides / Autoskill

Autoskill

Identify repeated research workflows and turn them into reviewable skill proposals.

Autoskill examines a user-requested window of recorded computer activity to look for recurring patterns of work. It matches those patterns against existing scientific skills and can propose reuse, a composition recipe, or a new draft. This is a workflow-discovery aid: it produces suggestions for review rather than evidence that a researcher performed or validated a particular analysis.

[Image: An explicitly selected activity window is redacted, grouped into recurring workflows, and matched to skill proposals.]

An explicitly selected activity window is redacted, grouped into recurring workflows, and matched to skill proposals. [View the full-size workflow diagram](../images/autoskill.png).

Questions this skill can help you explore

  • Which research tasks am I repeating? Group similar activity sessions within a chosen history window.
  • Does an existing skill already help? Compare recurring workflows with the installed skill descriptions.
  • What should a reusable workflow include? Review proposed drafts or recipes before promoting them.

What you bring

Provide an explicit request to analyze a defined activity window, access to a running Screenpipe daemon, and a directory of candidate skills. Choose the model backend and review its data destination. A representative history window and clear task boundaries help distinguish repeated research work from coincidental use of the same applications.

How the workflow works

  1. Check configuration and connectivity. Confirm the selected history source, skills directory, and language-model backend.
  2. Retrieve the requested window. Fetch activity with pagination and reject incomplete retrieval rather than silently summarizing a partial record.
  3. Redact and group. Scrub recognizable secret patterns, segment sessions, and cluster recurring application sequences.
  4. Match and propose. Use local embeddings to find relevant skills, then ask the configured model to suggest reuse, composition, or a new draft.
  5. Review before adoption. Inspect the report and drafts, correct their scientific content, and validate proposed skills before promotion.

What you get

OutputWhat it helps you do
Activity-cluster plan or reportReview which repeated patterns the system detected.
Matched skill suggestionsFind existing procedures relevant to repeated work.
Draft skills or composition recipesProvide editable starting points for a reusable workflow.

Example request

Use Autoskill to inspect the research activity I recorded during this explicitly selected week. Start with a dry run so I can review the recurring workflow clusters. Then use my configured local model to propose existing skills or draft procedures for the repeated tasks, keeping all proposals separate from the installed collection.

This is an illustrative request, not a reported research result.

Interpreting the results

Recorded activity is incomplete evidence of scientific work. Application sequences and window titles may miss the actual reasoning or outcome, and similarity scores do not establish that a skill is appropriate. Proposed procedures need domain review and ordinary validation.

Redaction is a safeguard rather than a guarantee. Detection and embedding inference run locally, but the selected model receives redacted summaries and matched descriptions; a remote backend sends that information off the machine.

Get started

The workflow requires Python 3.10+, httpx, PyYAML, sentence-transformers, Screenpipe, and a configured language-model endpoint. Local LM Studio is the default route; cloud backends are opt-in and require credentials. Initial model downloads need network access. Authentication may also be required by the local services.

Setup and technical instructions

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

Repository stars
47,290
Repository forks
4,282
7-day trend score
54

Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/k-dense-ai/scientific-agent-skills/skill-guides/badge)](https://verified-skill.com/skills/k-dense-ai/scientific-agent-skills/skill-guides)

Something wrong with this skill or its page? Tell us.