skills

In the author's own words, from each skill's SKILL.md — showing the first 25 of 49 files.

• huh

  Decode a message that didn't land: restate it in plain sentences, expand invented shorthand, split report from request, and flag claims stated without evidence. Use when the user signals your last message didn't land ("huh?", "what do you mean"), or pastes agent output from another session to translate.

  skills/author/huh/SKILL.md on GitHub (opens in a new tab)

• plainspoken

  Write like a person in everything you emit. Plain words, concrete claims, no AI tells. Use when writing any chat reply, status update, close-out, review comment, or commit message (always on for those), when a passage you drafted reads generated, or when reviewing prose another agent wrote.

  skills/author/plainspoken/SKILL.md on GitHub (opens in a new tab)

• writing-for-agents

  Write and prune documents an agent consumes: a SKILL.md, an AGENTS.md or CLAUDE.md, a rules file, a reference file reached by a pointer. Use when authoring or revising a skill, editing standing agent instructions or recording a decision as a rule agents will follow (a CLAUDE.md, AGENTS.md, CODE-RULES, or conduct-rules entry), or diagnosing why a document fires unreliably or gets skimmed.

  skills/author/writing-for-agents/SKILL.md on GitHub (opens in a new tab)

• writing-for-humans

  Write copy a human reads: a landing page, a README's front half, a launch post, a profile page. Use when drafting or revising public-facing prose, when a page reads like agent documentation, or when a finished draft needs its last-mile scrub for AI tells.

  skills/author/writing-for-humans/SKILL.md on GitHub (opens in a new tab)

• advisory-board

  Convene a multi-model advisory board: subscription-backed Claude, Codex, Gemini, and Grok CLIs reviewing the same material in formal, roundtable, or competitive mode. Use when the user asks for an advisory board, roundtable, panel, or idea tournament; a multi-model or cross-provider review or debate; a red-team of a plan, design, decision, or document by several models; or a consensus handoff fro…

  skills/decide/advisory-board/SKILL.md on GitHub (opens in a new tab)

• decision-map

  Chart or work a decision map for genuinely foggy work before any build slice is written: a new primitive, a milestone charter, an integration nobody can spec in one sitting. Use when asked to chart a decision map, work a map ticket, or plan a decision-heavy effort whose open questions gate each other.

  skills/decide/decision-map/SKILL.md on GitHub (opens in a new tab)

• grilling

  Interview the decider relentlessly to turn a half-formed plan, design, or idea into shared understanding: a design tree worked in rounds until nothing is silently assumed. Use when asked to grill, stress-test, or pressure-test thinking, and before charting a decision map or writing a spec off a conversation nobody has pressure-tested.

  skills/decide/grilling/SKILL.md on GitHub (opens in a new tab)

• computer-use

  Hand a GUI task on this Mac (click, type, read a window, move through an app) to Codex Computer Use, one bounded subtask at a time, and get back a result the runner has checked against what actually happened. Use when a task needs a desktop app driven or read and no API, CLI, or connector covers it, or when the user says use the computer, drive the app, or computer use.

  skills/in-progress/computer-use/SKILL.md on GitHub (opens in a new tab)

• fit-audit

  Rank this catalog's skills by what each would have saved this specific adopter, every top pick citing a real incident from their own history, and for a team judge each skill against the incumbent stack as well. Use when someone is considering the catalog and has not run setup, when the user asks which skills would help them most, when recommending a starting subset to a new adopter, or when audit…

  skills/in-progress/fit-audit/SKILL.md on GitHub (opens in a new tab)

• codebase-review

  Run a state review of the codebase: the counterpart to adversarial-review's change review. Use before building against an upcoming spec (make the change easy first), when lanes merged since the last review cross the repo's threshold, or when lanes or the orchestrator report the code fighting them.

  skills/investigate/codebase-review/SKILL.md on GitHub (opens in a new tab)

• ingest

  Turn a video, recording, voice memo, or media URL into an evidence packet of authoritative transcript, timestamped frames, and manifest, plus a routing recommendation. Use when the user shares a media file or URL with a goal, asks to process or transcribe a recording, or wants takeaways from a call, demo, playtest, or video.

  skills/investigate/ingest/SKILL.md on GitHub (opens in a new tab)

• prototype

  Build throwaway prototype code that answers a design question: UI variants on a live route, or a terminal UI or single shareable HTML file over a pure logic module. Use when a question is "how should this look / behave / feel in action" and discussion or static artifacts cannot settle it.

  skills/investigate/prototype/SKILL.md on GitHub (opens in a new tab)

• research

  Run an autonomous research lane against primary sources, ending in cited findings and, when the missing facts are human-held, a questionnaire. Use for investigations of vendor materials, regulations, upstream repos, recorded calls, or any fact-finding that should run fire-and-report.

  skills/investigate/research/SKILL.md on GitHub (opens in a new tab)

• domain-memory

  Keep per-repo institutional memory: a domain glossary plus lightweight decision records, written as side effects of work. Use when a grilling closes, a review rejection or declined finding is dispositioned, the decider corrects a session's wrong assumption, a session starts in a repo whose binding names a memory home, or the store passes its size bound.

  skills/orient/domain-memory/SKILL.md on GitHub (opens in a new tab)

• router

  Entry point for the team-workflow pack: names every pack skill and when to reach for it. Use when unsure which team-workflow skill applies, or to orient a new session/repo on what the pack offers.

  skills/orient/router/SKILL.md on GitHub (opens in a new tab)

• setup

  Once-per-repo binding interview for the team-workflow pack: records how the pack composes with this repo and seeds the binding doc the other skills read. Use when installing the pack into a repo, refreshing an existing installation's bindings, or auditing a consuming repo for drift after a pack release (audit mode).

  skills/orient/setup/SKILL.md on GitHub (opens in a new tab)

• adversarial-review

  Run an adversarial review of a change before it ships: isolated finders trying to break it, a skeptic pass that kills unproven findings, and a gate on confirmed blockers. Use before committing substantial work, at lane close-out, when the user asks for an adversarial or red-team review of a diff/branch/PR, or before external reviewers see the change.

  skills/run/adversarial-review/SKILL.md on GitHub (opens in a new tab)

• blast-radius

  Find what a change breaks somewhere else before it ships, beyond the diff and past where grep stops, and prove the one fact it is safe because of by running real code. Use when asked for the blast radius of a change, when deciding whether an in-flight change is safe to merge, or when reviewing a small diff you do not trust yet.

  skills/run/blast-radius/SKILL.md on GitHub (opens in a new tab)

• diagnose

  The disciplined bug-fixing loop: no fix ships without a cause the fixer can state in one plain sentence, with evidence. Use when any fix is about to ship without a nameable cause, when a bug resists its first fix attempt, or when a bug-shaped lane brief points here.

  skills/run/diagnose/SKILL.md on GitHub (opens in a new tab)

• handoff

  Write a structured session handoff so a fresh session resumes losslessly. Use when context is filling up (around half the window), when wrapping a work session, or when the user says "handoff", "checkpoint", "wrap up", or "save state".

  skills/run/handoff/SKILL.md on GitHub (opens in a new tab)

• implement

  How a working lane builds an item: seam-scoped test-first, tracer-first sequencing, a green checkpoint per slice, and file-don't-fix scope discipline. Use when a build-shaped lane brief points here, when implementing a spec or tracked item, or when mid-build discoveries tempt a lane beyond its brief.

  skills/run/implement/SKILL.md on GitHub (opens in a new tab)

• orchestrate

  Run one session as the lead developer of an agent team: it takes tracked work, routes it into parallel lanes, audits what comes back, and owns the merge, so the human brings decisions and never implements. Use when the user says "orchestration mode", asks one session to coordinate several lanes/sessions, or hands over an orchestrator role.

  skills/run/orchestrate/SKILL.md on GitHub (opens in a new tab)

• show-me-your-work

  Keep a reviewable decision trail for long-running or unattended work: one append-only TSV log, one row per decision with its why, evidence, and result. Use when starting an autonomous, multi-phase, or loop run a human reviews after stepping away, when a reviewer needs the trail to trust a result without rerunning it, or when the user asks for a decision log or says show your work.

  skills/run/show-me-your-work/SKILL.md on GitHub (opens in a new tab)

• to-tickets

  Turn a plan, a spec, a closed decision map, or a pressure-tested conversation into tracer-bullet work items on the repo's tracker, each an issue-as-spec with its blocking edges wired. Use when asked to file tickets, break work down, or move from "we've decided" to "it's on the board".

  skills/run/to-tickets/SKILL.md on GitHub (opens in a new tab)

• wizard

  Generate an interactive bash wizard that walks a human through a procedure only they can perform: third-party dashboards, credentials, DNS records, CI secrets, one-off cutovers. Use when a task stalls on steps an agent cannot take, or when a manual setup is tedious to re-explain every time. Not for steps the agent can perform itself.

  skills/run/wizard/SKILL.md on GitHub (opens in a new tab)

No skill matches that filter.

Similar skills

• ponytailby DietrichGebert Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote. 164k stars · 6 skills
• marketingskillsby coreyhaines31 Marketing skills for Claude Code and AI agents. CRO, copywriting, SEO, analytics, and growth engineering. 54.9k stars · 30 skills
• khazix-skillsby KKKKhazix 数字生命卡兹克开源的 AI Skills 合集 | Agent Skills: leader（帮你定义目标）, neat-freak 洁癖, hv-analysis, khazix-writer & more — Claude Code, Codex & 40+ agents 21.4k stars · 6 skills
• agent-skills-platformby FrancyJGLisboa Build tested agent skills and govern their lifecycle through a user-defined marketplace: evidence, discovery, updates, rollback, quarantine… 2.4k stars · 6 skills
• open-stepsby kharmanskyi Plain-language agent skills for Claude Code, Codex, Cursor and Gemini CLI: honest session reports, straight verdicts, steps you can follow.… 1.4k stars · 8 skills
• ECCby affaan-m The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Cod… 280k stars · 30 skills