All categories
AI & Agents
LLMs, agent workflows, RAG, MCP, prompting, and AI app patterns
- 300,204
- 12,509
Security grades appear on each card once the skill has been scanned. Newly imported skills may briefly show without a grade until the backfill job runs.
Open in full browserBrowse ai & agents skills
Showing 361–384 of 300,204 skills
- Masterdeck NotesWrite to the user's MasterDeck notes (the Notes panel in MasterDeck) — a new note, more text on a note you made, or the note of a ticket. Use when the user asks to "write it to a note", "save this in notes", "add to the ticket's note", "make a note of …", or wants something kept in MasterDeck's notes.Votes: 0GitHub stars: 3
- MasterUse when the user types /master, asks what work is pending across their Claude sessions, wants issues/PRs/meeting action items delegated to sessions, asks to message or spawn a session for an issue, when a /loop sweep fires, or when a `master watch` Monitor event arrives. Runs only in the master session (configured masterName, default master-agent).Votes: 0GitHub stars: 3
- Kill WorktreeTear down the session's worktree while preserving all work — stash uncommitted changes, keep the branch, return the session to the main repo, check the branch out there, pop the stash, and delete the worktree directory. Use when the user says "/kill-worktree", "drop the worktree", or wants worktree work moved back to the main checkout.Votes: 0GitHub stars: 3
- Babysit WorktreeCreate an isolated git worktree and switch this session into it so all further changes happen there, keeping the main checkout clean. Use when the user says "/babysit-worktree", "work in a worktree", or wants this session's changes isolated from the main working copy. Pair with kill-worktree to fold the work back.Votes: 0GitHub stars: 3
- Babysit TicketUse when the user types /babysit-ticket, asks which ticket/task this session is for, asks to link the session to a GitHub issue or the configured project board, or asks to move a ticket's board status (e.g. In Dev, PR Raised, Dev Done). Also use when a babysit-ticket hook message says the session is not linked.Votes: 0GitHub stars: 3
- Babysit PrPre-PR self-review plus post-create review-comment babysitting. Use before creating a PR (review the branch diff, fix findings, write the self-review marker) and after creating one (poll review threads, fix, reply, resolve, until merged or all threads resolved). Optional: MasterDeck has its own PR watch and self-review gate, so run it by hand (/babysit-pr) for the deeper pass, or wire its hooks per workspace in .claude/settings.local.json.Votes: 0GitHub stars: 3
- MailRead, search, file and answer the user's mailboxes through the claude-acc mail gateway (MCP server `mail`). Use when a task needs an email - a reply from someone (registrar, AWS, support, client), a verification code or link, an attachment, cleaning up an inbox, drafting or sending an answer, or waiting until a reply arrives.Votes: 0GitHub stars: 2
- DesktopDrive the whole macOS desktop with Anthropic's computer use toolset through the claude-acc desktop gateway (MCP server `desktop`): screenshot the active display, click, type, scroll and read small text with zoom, the same way the `browser` gateway drives the user's Chrome. Use when a task needs a native app or the desktop itself - opening or operating an app (Finder, TextEdit, Preview, System Settings, a native client), a menu, a window, a dialog, the Dock or menu bar, dragging between apps, ...Votes: 0GitHub stars: 2
- paper-proofread한국어 학술 논문 교정교열 스킬. 논문 .docx/.pdf/.md 파일을 입력받아 맞춤법, 띄어쓰기, 문장 교열, 논리 구조, 인용·참고문헌까지 체계적으로 점검하고 수정 원고와 교정교열표를 출력한다. 다음 상황에서 반드시 이 스킬을 사용한다: '교정', '교열', '교정교열', 'proofreading', '논문 검토', '논문 수정', '원고 다듬기', '맞춤법 검사', '투고 전 검토', '문장 교열', '학술 글쓰기 점검' 등의 키워드가 포함된 경우. 논문이 아닌 일반 문서(보고서·제안서 등)의 교정교열 요청에도 사용 가능하다. 단순 번역이나 요약 요청에는 사용하지 않는다.Votes: 0GitHub stars: 6
- Writing PlansALWAYS loaded FIRST when starting any new task — non-negotiable, before any tool use, regardless of whether the task seems trivial. Triggers on EVERY user turn: questions, requests, code, debugging, design, build, create, make, write, fix, add, change, set up, plan, implement, refactor, install, deploy, schema, SQL, database, YAML, JSON, config, API, function, class, test, document, hello, help me. Workflow: (1) If other skills apply (domain, language, framework, integration), load them FIRST...Votes: 0GitHub stars: 26
- Update Project State GraphUse when a plan that touched a registered project reaches the NEEDS_REVIEW state, when reviewing plan completion for a project-state change, or when checking whether a git diff broke consistency with a project's deep state graph. Keywords - NEEDS_REVIEW, project state review, maintain project state, stale reference check, renamed function still called, code graph consistency, agent review close, review plan completion.Votes: 0GitHub stars: 26
- Test Driven DevelopmentMUST BE ACTIVATED BEFORE EVERY CODING TASK — NO EXCEPTIONS. Applies to all production code, schema and config work: features, bugfixes, refactors, functions, classes, SQL schemas, migrations, YAML/JSON configs, API endpoints, CLI tools, scripts, handlers, parsers, validators, lint rules — anything that runs or gets validated. Iron law: NO PRODUCTION CODE WITHOUT A FAILING TEST FIRST. RED → GREEN → REFACTOR; code written before its test gets deleted. No qualifier exempts you: 'simple', 'trivia...Votes: 0GitHub stars: 26
- Systematic DebuggingUse when encountering any bug, test failure, or unexpected behavior, before proposing fixesVotes: 0GitHub stars: 26
- Subagent Driven DevelopmentUse ANY time work should be delegated to sub-agents (Agent tool) instead of done in the main session: (a) executing a plan task-by-task, a fresh sub-agent per task with two-stage review; (b) parallel dispatch for 2+ independent problem domains; (c) routing a task to the RIGHT agent type (Explore, Plan, a project or plugin agent) instead of a generic one; (d) one-shot huge reads (giant logs, full API references, monorepo scans) handed to one sub-agent that returns only a summary. CRITICAL TRIG...Votes: 0GitHub stars: 26
- ReceiptsHelp the user reply to a colleague who questioned a decision — read-only. Use when the user types `/receipts <what they said>`, or asks how to answer someone who challenged a number, a change or a choice in this project (\"why did you change this\", \"didn't we agree the other way\", \"where did this number come from\"). Writes a reply grounded ONLY in recorded rows, lists the evidence under it with a record id per line, and names what the user should confirm or add before sending. Never send...Votes: 0GitHub stars: 26
- Project State GraphUse when setting up a new project, creating a new app, kicking off a new initiative, onboarding an existing repo, or whenever you need a repo graph / code graph / project state graph / map of a codebase before working in it. Keywords - set up project, new project, new app, new initiative, project state graph, repo graph, code graph, codebase map, architecture overview.Votes: 0GitHub stars: 26
- LedgerAsk the ledger why / whether we tried — read-only. Use when the user types `/ledger <question>` or asks why something in this project is the way it is, why a value was chosen, who decided it, whether an alternative was ever tried, or whether a change was ever verified. Answers ONLY from recorded rows: every sentence cites the record id it rests on, absences are computed by code, and the searched range is stated. Changes nothing the project depends on.Votes: 0GitHub stars: 26
- Executing PlansUse IMMEDIATELY after writing-plans, OR whenever a plan exists in ~/skill-workspace/orchestrator.db. Owns ALL post-publish writes to the orchestrator database via the deterministic-flow scripts in scripts/ (run-step is the preferred path for shell work). Never bypass these scripts; never write to the DB ad-hoc. Triggers on every plan execution: code, SQL, schema, build, test, fix, implement, deploy, refactor, run, execute, continue, resume. When a test fails: never patch the test silently — f...Votes: 0GitHub stars: 26
- BrainstormingYou MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.Votes: 0GitHub stars: 26
- Kdna LoaderRead a user-selected local .kdna asset through the exact official CLI/Core/Read or an operator-bound local MCP process. Never discover assets or treat model-supplied paths as permission.Votes: 0GitHub stars: 3
- Kdna CreatorAuthor and revise KDNA assets from authorized materials, review their substantive content and field coverage, and use the Host-selected native CLI or the explicitly matched Studio creation session.Votes: 0GitHub stars: 3
- Prepare For WishesEntry point skill that loads all global guidelines, analyzes the project's technical context (if a repo exists), announces readiness, and asks what the developer wants to do — redirects to list-tasks, analyze-task, or implement-task as needed.Votes: 0GitHub stars: 2
- List TasksBrowse available tasks (user story or bug) from the project's docs/ folder. Lists them for selection, then hands off to analyze-task for deeper validation.Votes: 0GitHub stars: 2
- Implement TaskImplement a validated task end-to-end. Checks that analyze-task ran within the last hour, then branches, writes tests (TDD), implements, commits, lints, and commits again.Votes: 0GitHub stars: 2