Skip to content
A/Agent Skills
← Back to catalog

Selection guide / Writing

Know the job.
Choose the tool.

Use one lead skill for the main task. Add a specialist only when it owns a separate phase.

Writing and Knowledge Skills

Article pipeline

  1. $writing-fragments mines useful raw material without forcing structure.
  2. $writing-beats turns material into a narrative journey.
  3. $writing-shape builds the article paragraph by paragraph.
  4. $edit-article restructures and tightens an existing draft.

Start at the stage your material is actually in; a complete draft can go directly to $edit-article.

Knowledge and teaching

  • $obsidian-vault: find, create, and organize linked notes in Obsidian.
  • $teach: teach a concept or skill inside the current workspace.
  • $scaffold-exercises: create structured course exercise directories.
  • $writing-great-skills: guidance for authoring predictable agent skills.
  • $research: delegate reading legwork to a background agent that checks high-trust primary sources and leaves the findings as Markdown in the repo. Use it when the answer must be sourced rather than recalled; not for questions the codebase itself answers, which $graphify handles.

Writing skills change the shape of content. They are not substitutes for domain research or factual verification — $research is the skill that does the verifying.

README star conversion

A bundle that scores and improves a GitHub README against patterns from top-starred repos, aimed at turning visitors into stars.

  • $readme-roast: the orchestrator — run it for a full audit that delegates to the scorers below and aggregates a weighted 0-100 score with an action plan. Choose it when you want the whole picture; reach for a single scorer only when you already know which dimension to fix.
  • $readme-audit: the same full parallel audit surfaced directly, when you want the composite report without the roast framing.
  • $readme-star-killers: lightweight diagnostic of only the highest-impact gaps — use when you want the short list, not the full report.
  • Single-dimension scorers: $readme-hero (value prop), $readme-install (quickstart friction), $readme-trust (badges, social proof), $readme-visuals (GIFs, screenshots), $readme-structure (scannability), $readme-benchmark (comparison against category leaders). Use one when a full audit already told you which dimension is weak.
  • Output and iteration: $readme-rewrite drafts improved sections from audit findings, $readme-report and $readme-report-pdf produce shareable reports, $readme-compare measures before/after deltas, $readme-history tracks progression over time.

Use this bundle only for README-to-star conversion of a GitHub project. For general documentation or article shaping, use the article pipeline above; for authoring an agent skill, use $writing-great-skills.