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

75%

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

Fix and improve this skill with Tessl

tessl review fix ./skills/autoskill/SKILL.md

Quality

Content

65%

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

The body is highly actionable with concrete commands and a clear pipeline, but it is somewhat verbose in its prerequisites and has dangling references to a missing config.yaml and tests/ directory. Workflow clarity is good but lacks explicit mid-pipeline validation and fix-and-retry feedback loops.

Suggestions

Ship the referenced config.yaml (and tests/ suite) in the bundle, or stop pointing to them — every run/promote command passes --config config.yaml yet no such file exists.

Add an explicit fix-and-retry loop after `doctor` (e.g. "if doctor exits non-zero, fix the marked error and re-run until it passes before proceeding to run") and a mid-pipeline checkpoint note.

Trim the from-source build and LM Studio setup detail into a references/setup.md, keeping the body to the doctor/run/promote essentials.

Dimension	Reasoning	Score
Conciseness	Mostly concrete and operational, but the prerequisites section pads with build-from-source detail (cargo build, macOS Xcode troubleshooting, LM Studio GUI-vs-CLI notes) and mild marketing prose ("strong reasoning at a size that fits on most workstation GPUs") that could be tightened.	2 / 3
Actionability	Provides fully executable, copy-paste-ready commands throughout (doctor/run/promote invocations, `export SCREENPIPE_TOKEN=$(screenpipe auth token)`, `pipenv install ...`, `lms load ...`), a complete config.yaml block, and a concrete architecture diagram naming each script.	3 / 3
Workflow Clarity	The workflow is well sequenced (doctor → run → promote) with a preflight and --dry-run gate, but the core run pipeline is described as straight-through with no mid-step checkpoints, and recovery actions on doctor/promote failures are implied rather than spelled out as fix-and-retry loops.	2 / 3
Progressive Disclosure	Existing references are one-level-deep and clearly signaled (references/screenpipe-config.yaml, references/https-proxy.md, scripts/*.py all present), but the body repeatedly points to `config.yaml` and `tests/` which do not exist in the bundle, and a fair amount of setup content is inlined rather than split out.	2 / 3
	Total	9 / 12 Passed

Description

85%

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 specific, complete, and distinctive, using third-person voice throughout with an explicit Use-when trigger. Its only weakness is trigger-term breadth: it relies on one natural phrasing rather than enumerating the common variations a user might say.

Suggestions

Broaden the trigger clause to enumerate natural variations, e.g. "Use when the user asks to analyze their recent work, draft a skill from their workflow, find what's not yet covered, or propose composition recipes for repeated workflows."

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...)" — matching the anchor for several specific concrete actions.	3 / 3
Completeness	Explicitly answers what (observe/detect/match/draft) and when via an explicit "Use when..." clause, satisfying the anchor for clearly answering both what and when.	3 / 3
Trigger Term Quality	The single "Use when the user asks to analyze their recent work and propose skills based on what they actually do" is a natural trigger, but it offers one composite phrasing rather than the broader coverage of variations (e.g. "draft a skill", "find composition recipes", "what's not covered") that appear only in the body.	2 / 3
Distinctiveness Conflict Risk	The screenpipe-driven workflow-observation-to-skill-proposal niche with triggers like "analyze their recent work and propose skills" is distinct and unlikely to fire for unrelated skills.	3 / 3
	Total	11 / 12 Passed

Validation

87%

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 — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_field	'metadata' should map string keys to string values	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	14 / 16 Passed

Repository: K-Dense-AI/scientific-agent-skills
Path: skills/autoskill/SKILL.md
Commit: 757b63b

Reviewed: about 11 hours 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.