autoskill

Observe the user's screen via screenpipe, detect repeated research workflows, match them against existing scientific-agent-skills, and draft new skills (or composition recipes that chain existing ones) for the patterns not yet covered. Use when the user asks to analyze their recent work and propose skills based on what they actually do. Requires the screenpipe daemon (https://github.com/screenpipe/screenpipe) running locally on port 3030 — the skill has no other data source and will refuse to run if screenpipe is unreachable. All detection runs locally; only redacted cluster summaries reach the LLM.

Quality

83%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Quality

Content

75%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is a well-structured, largely executable overview with a clearly sequenced multi-step pipeline and validation checkpoints, supported by real bundle files for the detail it delegates. Its main weaknesses are a missing config.yaml artifact (hurts actionability and navigation) and slightly over-explained install/citation passages.

Suggestions

Ship the referenced config.yaml inside the bundle (or point the examples at references/screenpipe-config.yaml) so the run/doctor commands are fully copy-paste-ready and no referenced path dangles.

Trim the screenpipe from-source build walkthrough and the citation block to the essentials, or move them into a references/ file to keep SKILL.md a lean overview.

Add an explicit fix-and-retry feedback loop after the doctor step (e.g., 'if any check is `error`, resolve the listed item and re-run doctor') to close the workflow-clarity gap.

Dimension	Reasoning	Score
Conciseness	Mostly efficient with lean, copy-paste command blocks and an architecture diagram rather than padded prose, but a few sections (detailed screenpipe-from-source install steps, the citation block) are over-explanation relative to the core task; not 5 because of these trimmable passages, not 3 because the bulk is genuinely lean.	4 / 5
Actionability	Provides concrete executable commands throughout (export token, cargo build, lms load, the autoskill.py run/doctor/promote invocations) and a step-by-step internal pipeline, but the repeatedly-referenced config.yaml is absent from the bundle, leaving a gap in fully copy-paste-ready execution; not 5 because of that missing artifact, not 3 because guidance is largely executable.	4 / 5
Workflow Clarity	The fetch→redact→cluster→match→synthesize→report pipeline is clearly sequenced with a doctor preflight and a --dry-run/--plan gate before the LLM call, and promote refuses overwrite; not 5 because explicit error-recovery feedback loops (fix-and-retry on doctor failure) are implicit rather than spelled out.	4 / 5
Progressive Disclosure	SKILL.md functions as an overview delegating detail to one-level-deep, clearly signaled bundle files (references/https-proxy.md, references/screenpipe-config.yaml, scripts/*.py) with well-organized sections; not 5 because config.yaml and tests/autoskill are referenced but not present in the bundle, leaving dangling navigation targets.	4 / 5
	Total	16 / 20 Passed

Description

92%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is concrete, complete, and clearly niche-scoped, with an explicit "Use when" trigger and a hard dependency/refusal condition that further distinguishes it. The only weakness is trigger-term variety, which is good but not exhaustive of the synonyms a user might naturally say.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — "Observe the user's screen via screenpipe, detect repeated research workflows, match them against existing scientific-agent-skills, and draft new skills (or composition recipes that chain existing ones)" — giving comprehensive coverage; not 4 because there are no meaningful coverage gaps.	5 / 5
Completeness	Explicitly answers both what (observe/detect/match/draft) and when (a clear "Use when..." clause with concrete trigger phrasing), matching the anchor for an explicit what-and-when with triggers; not below 5 because both halves are concrete.	5 / 5
Trigger Term Quality	Includes a natural user phrase — "Use when the user asks to analyze their recent work and propose skills based on what they actually do" — but the trigger language is somewhat formal and does not exhaustively enumerate synonyms/variations a user might say; not 5 because coverage is good rather than comprehensive.	4 / 5
Distinctiveness Conflict Risk	Occupies a clear niche (screenpipe-driven generation of scientific-agent-skills) with distinct triggers and an explicit refusal condition, so overlap with sibling skills is minimal; not 4 because the niche is unambiguous.	5 / 5
	Total	19 / 20 Passed

Validation

93%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_field	'metadata' should map string keys to string values	Warning

	Total	15 / 16 Passed

Repository: K-Dense-AI/scientific-agent-skills
Path: skills/autoskill/SKILL.md
Commit: 49c6e97

Reviewed: 9 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.