Browse reusable Agent Skills, each with a clear purpose and practical guidance.
【CAPABILITY SKILL】基于用户提供的参考文案样本，对视频素材内容进行深度文风仿写，生成风格化脚本。Based on user-provided reference text samples, the video material is deeply rewritten in terms of writing style to generate a stylized script.
Product Manager (Morgan). Use for PRD creation (greenfield and brownfield), epic creation and management, product strategy and vision, feature prioritization (MoSCoW, RICE), roadmap planning, business case development, go/n... Triggers: PRD, create epic, execute-epic, roadmap, product strategy, pm, @pm, requirements...
Copywriting / prose / tone style guide for EVERY user-facing or public-facing text artifact. INVOKE THIS SKILL (via the Skill tool) BEFORE you write or edit ANY of the following — and if you are a SUB-AGENT instructed to do any of them, invoke it yourself first; the agent that writes the copy loads this skill, the orchestrator does not load it on a sub-agent's behalf: (1) ANY GitHub-facing text — before running `gh issue comment`, `gh pr comment`, `gh pr create` / any `--body`, `gh issue close --comment`, `gh pr review` / `gh pr comment` review notes, or `gh release create` / `gh release edit --notes`; i.e. any issue reply or comment, PR description/body, issue-close note, code-review comment, or release notes. (2) ANY documentation — creating or editing a file under `site/content/docs/**` or `site/content/blog/**`, `README.md`, `CHANGELOG`, `wiki/**` user-facing pages, or any `.md` / `.mdx` that ships to users. (3) Marketing / homepage / blog copy, a `package.json` or npm `description`, or any other description/summary field. (4) Whenever you APPLY a general copy-style correction — sweep it everywhere it applies, not just the one spot it was raised, and record it in PROSE.md. (5) A substantive prose chat reply. (6) Reformatting or structuring existing copy for scannability — breaking up a wall of text, or converting a run of paragraphs into a list, table, or callout. A bold-sentence lead-in does NOT count as a block-level break: a run of bold-led paragraphs is still a wall of text, and the fix is a real list/table/ callout, never bolding the first sentence of each paragraph. Through-line: factual, neutral, terse, scannable — never two-plus dense paragraphs in a row without a block-level element; state what's true, cut everything that doesn't add a fact, build for a reader who skims. Encodes GitHub maintainer-hygiene tone, sentence/heading mechanics (never open a sentence or heading with inline code), scannability, honesty/restraint, release-notes + marquee-announcement + migration-entry shape, markdown mechanics (never hard-wrap paragraphs), and the universal tone rules. Canonical full guide: PROSE.md at the repo root (this skill is its trigger + index). Project-specific copy rules (brand vocabulary, claim-tracing, product framing) live in AGENTS.md, layered on top.
Write effective rebuttals to reviewer comments for journal submissions
Remove AI-generated patterns to produce natural, authentic academic writing
Systematic approaches to curriculum design using backward design and alignment
Use grammar and style checking tools to polish academic manuscripts
Guide to writing effective research paper introductions
Structured framework for writing peer review reports and paper critiques
Craft effective point-by-point reviewer response letters