Productivity

Guided spec-driven development (SDD). Use when the user starts a feature, sends or pastes a task.md / requirement / idea to implement, asks "what's next" in an SDD flow, or wants to write a spec, design, tasks or review document, or set up docs/ for SDD. Triggers on "nova feature", "crie uma spec", "sdd", "docs/specs", "task.md", "próximo passo", "continuar", "o que falta".

End-of-session audit and safe cleanup of a git repo and the machine. Finds leftovers (untracked junk, unpushed commits, stale or dirty worktrees, merged and unmerged branches, stashes, lane harness state, dev servers and background processes, temp and build artifacts, open tracker items, deploy drift), removes only what is provably stale and recoverable, and asks before anything that could lose work. Use whenever the user says "cleanup", "clean up", "does anything need to be cleaned up", "session cleanup", "tidy up", "end of session", "wrap up", "are we done here", or before writing a final session report, even if they do not name this skill. Works with or without the /lanes harness. Not for refactoring, lint fixes or dead-code removal.

Keeps the project notes of an Obsidian vault up to date with work done in Claude Code, respecting the vault's own structure and the hand-written prose of each note. Use whenever Obsidian, the vault, a project note or a canvas comes up: \"update the note\", \"document this\", \"sync the vault\", \"write this down\", \"where was I?\", or when a stretch of work ends and what changed should be recorded. Also to review which projects exist and in what state, add notes to a project's folder, update a canvas, and install or fix the hook that logs sessions. In Spanish: \"actualiza la nota\", \"documenta esto\", \"sincroniza\", \"apunta esto\", \"¿por dónde iba?\".

Monthly bookkeeping and payroll for a therapy private practice (solo or group). Use when the practice owner wants to (1) turn bank and credit-card statements into a filled monthly profit-and-loss workbook using a fixed chart of accounts, with every transaction classified, internal transfers excluded, and the month tied out to the penny; and/or (2) build and deploy a companion "Practice Books" web app that runs monthly contractor/employee payroll (1099 + W-2 estimates) and shows a read-only visual dashboard of the P&L (charts, trends, category mixes). Reference implementation: Google Sheets + Google Apps Script, but the workflows are written to work with any spreadsheet/app stack the agent can drive.

Use when a user asks to investigate a ticket, issue, bug, feature request, or technical change and produce an implementation plan before coding, especially when existing work, linked requirements, runtime evidence, or execution handoff must be reconciled.

Builds a sales or pipeline dashboard from a CRM opportunity export: open and weighted pipeline, win rate, stale deals, deals closing soon and pipeline by owner, using only the columns it needs and reconciled to the export. Use when asked for a pipeline or sales dashboard, or KPIs from a CRM or Salesforce export.

Builds Excel workbooks (.xlsx) other people can trust: real formulas, one table per sheet, validation where people type, and a README tab. Use when asked for a spreadsheet, tracker, grid or Excel report, or to turn data into a workbook.

Sets up a recurring multi-stage job (a quarterly business review, a monthly recap, a launch readout) as a workflow folder of numbered stages, each with a short contract saying what it takes in, what it hands on, which skill does the work and what a person checks; then runs it one stage at a time, stopping for that check. Use when asked to set up a workflow, a pipeline or a repeatable process, to break a recurring job into stages or steps, or to run the next stage of one. Not for making one deliverable on its own (use that deliverable's skill) or for setting up an account.

Builds a structured ASCII dependency graph (max 100 columns) for terminal output from a set of pending and settled decisions. Evaluates nodes across five fixed dimensions: sensible, recommended, user-typical (direction and decision-avatar confidence coded separately), coherent, and contradictory; edges carry relationship type and verification status. Relentlessly exposes blockers, unlocked paths, execution orders, and conflicting recommendations. Use for "which decisions depend on each other", "decision dependency graph", "visualize decision path", "check contradictions between decisions", "sequence of decisions", /decision-path. Do NOT use to first inventory decisions -- that''s decision-briefing; do NOT use to bundle the recommended overall concept for sign-off -- that''s decision-draw; do NOT use for compact five-part blocks with options and pro/contra -- that''s decision-shot; do NOT use to work out a single complex choice with frameworks -- that''s decide.

Condenses bundled pending decisions into ONE cohesive, prose-based overall concept assuming all recommended choices are adopted. Shows how subsystems interlock after implementation instead of forcing the user through dozens of isolated individual questions. The user accepts the picture as a whole or provides natural-language corrections, which are automatically translated back into individual decisions and written to the register via decision-briefing Phase 4. Use for "what does the whole thing look like if we agree to everything", "draw the big picture", "overall decision concept", "decision draw", /decision-draw, "how do the systems work together". Do NOT use when alternatives per item must be debated -- that's decision-briefing; do NOT use for dependency graphs -- that's decision-path; do NOT use for discrete blocks with options and pro/con breakdown (even for tightly grouped decisions) -- that's decision-shot; do NOT use to work out a single choice -- that's decide.

Use when the user is designing, reviewing or troubleshooting a Microsoft Copilot Studio agent that delegates to child agents, connected agents, external agents (A2A, Microsoft Foundry, Microsoft 365 Agents SDK), a live agent, or other agents through the Employee Self-Service agent, and needs to decide who may delegate which actions to whom, what context crosses the handoff, what comes back, and where confirmation, approval and credentials must sit. Trigger phrases include "connected agent permissions", "what does a connected agent see", "pass conversation history to this agent", "child agent context", "multi-agent handoff security", "can the subagent delete records", "A2A agent data sharing", "handoff to live agent variables", "Employee Self-Service handoff", "subagent replies twice", "decision boundaries for multi-agent". Do not use for single-agent tool design with no delegation, for non-Microsoft agent frameworks, or for shared agent memory governance.

Takes in the raw materials a project receives (briefs, chats and chat exports, letters, files, archives, screenshots, recordings, links to cloud documents) and keeps them in order in the project's inbox/, with passwords and personal data out of git and one inventory of what is there and what it says: who decided what and when, constraints, contradictions, questions to third parties. Use when the owner hands over materials or asks to sort them («разбери, что прислали», «вот всё от клиента, разбери», «разложи входящие», «что нового в inbox», «обнови опись»), and when other work, such as designing a feature, rests on incoming materials. Not for a single meeting recording, a register of open problems, or logs and errors. The owner may phrase this in any language; match on meaning.

Comprehensive Vivado project revision control strategies for standard, DFX (Partial Reconfiguration), IPI Block Design Container, and Segmented Configuration projects. Includes automated tools for project type detection, source analysis, settings capture, and build script generation. Use this skill whenever the user mentions version control, Git, build scripts, project portability, team collaboration, CI/CD for Vivado, exporting sources, preparing a project for handoff, or recreating projects — even if they don't explicitly say \"revision control\". Also trigger for \"make my project portable\", \"automate project recreation\", or \"set up Git for my FPGA design\".

Design — Assembles comprehensive board and investor update decks by pulling perspectives from all C-suite roles. Use

Turn a brief into a reviewed system design delivered as a clickable read-only HTML view and a PDF, with task and workflow diagrams, deterministic controls, risk analysis, evaluation fixtures and a milestone coding handoff. Use to detail AI workflows from minimal input; produces designs rather than implementing the business system.

Read and write cards on the user's self-hosted Slipdock boards with the `slipdock` CLI. Use whenever the user mentions their kanban or Slipdock board, cards, lists/columns, tags, flags, or asks to add, move, complete, flag, tag, check off, comment on, archive, or look up tasks/cards. Also use to summarise what is on a board, what is overdue, blocked, or in progress, to set up or inspect board automations (rules that email, move, flag or alert by themselves) and the alerts they raise, to connect runners (coding agents on the user's own machines that the board sends cards to) and follow or cancel their jobs, to send a meeting's recording or transcript, or a conversation with an AI assistant, for capture, to tidy a board up (the gardener's suggestions: epics done but open, stale or duplicate cards, orphaned pages), and to read or write the wiki pages, docs, runbooks, specs and decision notes kept on a board.

Đưa ảnh, video lên kho Cloudinary để có link công khai dùng được ngay, rồi ghi link vào bộ não để lần sau tìm lại. Trước khi đẩy lên sẽ kiểm có cần lên mây không, có bị trùng không, có lộ thông tin riêng không. Dùng khi người dùng đưa ảnh hoặc video và muốn cất làm nguyên liệu, kể cả khi chỉ nói "đẩy ảnh này lên kho media", "upload ảnh này", "lấy link ảnh này", "lưu video này để dùng sau", "đưa ảnh lên mây", hoặc hỏi tìm lại link ảnh đã lưu. Nội dung chữ thì dùng skill nap-vao-bo-nao.

Đưa nội dung từ bên ngoài vào kho bộ não thứ hai - đọc trọn nguồn, hỏi dùng vào việc gì, rút phần dùng được, đặt đúng chỗ và nối với trang sẵn có. Dùng khi người dùng đưa file, bài web, PDF, ghi chú, bản gỡ băng, ý tưởng và muốn giữ lại, kể cả khi chỉ nói "lưu cái này", "nạp vào kho", "ghi nhớ giúp tôi", "thêm vào bộ não", "xử lý file trong raw". Chỉ hỏi tra cứu thì trả lời trực tiếp. Ảnh, video cần link công khai thì dùng skill luu-media.

Turn one episode JSON file (show, title, summary, guests, links, chapters, transcript, duration_seconds) into an episode's publish-ready assets. `show-notes` renders the notes as Markdown or plain text; `chapters` validates the chapter list (first chapter at 0:00, starts strictly increasing, every start inside the duration, titles non-empty) and exports YouTube description chapters plus a Podcasting 2.0 chapters JSON — invalid chapters exit 1 and export NOTHING, because players silently ignore a broken list; YouTube's extra rules (>=3 chapters, each >=10s) are warnings, not problems. `audiogram-plan` ranks transcript segments by explained text signals (questions, numbers, exclamations, a fixed strong-phrase list, 15–60s length; floor 5, ties by earliest start) and prints a plan — it cuts no audio and renders no video. Use when finishing a podcast episode from an episode file. Exit 1 is a finding, not a failure.

Turn a meeting recording (video or audio file) into a transcript by speaker with screenshots of the shared screen and an optional summary, as Markdown, HTML and PDF. Use when the person gives a recording and asks to transcribe it, find who said what in it, or summarize it — «сделай транскрипт встречи», «расшифруй запись созвона», «саммари по записи». Not for a meeting that already has a text transcript.

Draft, schedule and publish social media posts, upload media, check engagement and best times to post, and answer inbox comments with the Sendcove MCP server (X, Instagram, TikTok, Threads, Bluesky, YouTube, Facebook Pages, LinkedIn). Use when the user asks to write, schedule, post or cross-post social content, attach an image or video to a post, review how posts performed, or read and reply to comments and mentions in Sendcove.

Especialista sênior em tráfego pago (Meta Ads, Google Ads, estratégia, mensuração, criativos, orçamento, auditoria e relatórios). Use SEMPRE que o usuário falar de anúncios, campanhas, mídia paga, CPA, CPL, ROAS, CPM, CTR, pixel, CAPI, GTM, UTM, Advantage+, PMax, Search, lances, escala, fadiga de criativo, público, segmentação, verba, auditoria de conta, relatório de performance, ou pedir uma opinião/decisão sobre tráfego — mesmo sem dizer "tráfego pago". Responde dúvidas direto e aciona as skills específicas (campaign-analyst, ads-copywriter, campaign-planner, budget-optimizer, campaign-direction, landing-page-trafego, client-research, plano-de-midia, rotina-trafego) quando a entrega é específica.

A sourced, dated field guide to YouTube, Instagram, TikTok, Facebook, LinkedIn, X and Threads. For each platform it gives formats and limits, how people find content, what it reports to creators, what the creator controls, and what nobody outside the company knows, with every fact labelled official, reported, inferred or study and dated. Use when someone asks how a platform works, what the algorithm rewards, which format to use there, or whether a rule they heard is true. Checks the facts database instead of repeating folklore.

Pre-launch audit for a website or SaaS - checks SEO tags (title, description, canonical, lang, Open Graph), robots.txt, sitemap.xml, favicon and apple-touch-icon, real 404 status, HTTPS redirect, legal pages (privacy, terms, refund), image alt text, page weight and response time, plus a payment self-check (webhook signature, raw body, idempotency, refunds, secrets, success-page grants), then lists the problems and fixes them in the codebase. Ships a no-dependency crawler script (audit_site.py) that produces a pass/warn/fail report. Use when the user says they are about to launch, ship, go live, submit to Product Hunt or AI directories, or asks "is my site ready", "SEO check", "why is my site not indexed", or wants a launch checklist, or asks whether their payment or webhook setup is safe.