polish-docs
Skill7 starsChanged 4 days agoMITView on GitHub
Polish README, onboarding, quickstart, documentation redesign. Improve first-use structure, persuasive evidence, and visual presentation; documentation-style covers ordinary prose edits.
In this repository22 other files it gives its agents.
- AGENTS.mdAGENTS.md
- audit-and-resolveSkill
- context-pagingSkill
- coordinate-workSkill
- design-skillSkill
- design-time-triggered-skillSkill
- design-validationSkill
- diagnose-and-repairSkill
- indefinite-loopSkill
- install-dev-skillsSkill
- operate-workflowSkill
- run-experimentSkill
- contract-disciplineSkill
- documentation-styleSkill
- evidence-disciplineSkill
- git-disciplineSkill
- scratch-spaceSkill
- testing-doctrineSkill
- context-compaction-procedureSkill
- maintain-scratchSkill
- refine-skillSkill
- security-hygiene-checkSkill
The file
We split this skill into the rules it gives an agent
9 rules. About 0 are common, 9 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈9
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 12
Treat README.md as human-facing, marketing-centric documentation: earn adoption through concrete benefits and evidence.
- Distinctive: line 14
Use scratch records only for unresolved documentation work or evidence needed to resume it.
- Distinctive: line 16
Use concrete, evidence-driven copy: no hand-waving, empty superlatives, or unsupported claims.
- Distinctive: line 18
Read the project and documentation to establish the product, intended reader, and concrete problem it addresses.
- Distinctive: line 32
Visual proof: use crisp screenshots, annotated diagrams, or short demonstrations of supported behavior.
- Distinctive: line 33
Consistency: use a small palette, one recognizable icon style, and consistent spacing and borders.
- Distinctive: line 34
Useful components: use icon cards for distinct routes, labeled code panels with copy controls where supported, and selective line …
- Distinctive: line 35
External grounding: when naming adopted conventions, link authoritative external explanations.
- Distinctive: line 36
Trust signals: place supported versions near installation, methodology beside benchmarks, and release status beside downloads.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Polish docs
Make the project's value and first-use path clear. Default to the README and directly related getting-started material.
Treat README.md as human-facing, marketing-centric documentation: earn adoption through concrete benefits and evidence. Put computer-facing, utility-centric routing and operating instructions in AGENTS.md; do not turn it into marketing copy.
Use scratch records only for unresolved documentation work or evidence needed to resume it. Keep settled conventions in their canonical documentation; do not log prose edits.
Use concrete, evidence-driven copy: no hand-waving, empty superlatives, or unsupported claims.
- Read the project and documentation to establish the product, intended reader, and concrete problem it addresses. Distinguish supported benefits from inferred promises.
- Compare introductions from successful projects with similar audiences. Choose relevant patterns to adapt; reference examples provide starting points.
- Make the opening answer what this is, why the reader would use it, and how to start. Name the product category and connect its mechanism to the reader’s problem; neither a slogan nor a capability list supplies that explanation. Support the benefit with concrete evidence and one primary next action.
- Follow the reader's adoption sequence: “Is this for me?” → “Why this?” → “Can I trust it?” → “How do I start?” Use a relevant example or demo as proof, address adoption barriers beside the call to action, then provide the shortest complete first-use path. When copy-paste prompts suffice, use them without an additional walkthrough. Link deeper reference and contributor material afterward.
- Apply a coherent visual treatment using the conventions below and the existing documentation platform.
- Read the page in order as a newcomer without conversation context. Check separately: can the reader identify the product, explain why it is useful, understand why each section is relevant, and take the next action? Do not count a feature description as a reason to use it. Verify claims, instructions, and rendered presentation as the task permits.
Use direct, reader-centered voice and the audience's terminology. Put proof beside each benefit; replace evaluative adjectives with observable behavior. Use descriptive headings and concrete skill or feature descriptions. Introduce unfamiliar categories by the need they address before listing members. Put a product-specific orientation sentence before installation prompts or choices; a shared introduction can serve both. Preserve this context when cutting repetition. For ongoing prose work, use documentation-style.
Visual conventions
Visual polish should make identity, behavior, and evidence easier to recognize and inspect.
- Opening: compose a compact wordmark, concrete value proposition, primary action, and real example or product image. Restrained gradients or illustration can establish identity without competing with the content.
- Visual proof: use crisp screenshots, annotated diagrams, or short demonstrations of supported behavior. Use consistent framing and evidence-specific captions; do not imply functionality with invented product imagery.
- Consistency: use a small palette, one recognizable icon style, and consistent spacing and borders. Keep reading surfaces quiet, with clear hierarchy and accessible contrast; reserve expressive decoration for the opening and transitions.
- Useful components: use icon cards for distinct routes, labeled code panels with copy controls where supported, and selective line highlighting. Keep commands and results distinct; use motion only to explain behavior or acknowledge interaction.
- External grounding: when naming adopted conventions, link authoritative external explanations. Describe the project’s adaptation accurately; internal files document implementation but do not independently substantiate the convention. Never imply endorsement or conformance beyond the evidence.
- Trust signals: place supported versions near installation, methodology beside benchmarks, and release status beside downloads. Link the license, changelog, and relevant limitations. Use a few meaningful badges, not a wall of implied endorsements.
For a GitHub README, favor a compact branded header, one useful visual, clean examples, and deliberate spacing. For a documentation site, add navigation cards, code controls, and theme support where the platform allows. Preserve readable text alternatives and narrow-screen usability.
Examples
Illustrative opening for a fictional tool; assume the stated capabilities are supported:
Find added, removed, and changed records in CSV exports.
Recurring exports can reorder rows, making a line-by-line comparison misleading. Rowdiff is a CSV comparison tool that matches records by customer ID so you can identify actual record changes.
Review record changes
The report includes changed records and omits unchanged ones. Comparison runs locally; no upload or account required.
Compare your first exports
Try the included sample files, then compare your own.
The example immediately following that invitation, after installation:
rowdiff previous.csv current.csv --key customer_idExample output:
Added: 3 records
Removed: 1 record
Changed: 7 recordsDoes it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

