Skill Composition
Each skill is designed for a specific kind of caller — some only fire when you name them, others load automatically when the model judges them relevant. The split below summarises both modes at a glance.
These skills only fire when you call them by name. The agent will not load them on its own — they stay inert until invoked.
These skills are loaded automatically by the agent when it judges the context relevant. You usually don't need to call them yourself.
Skill List
This skill enables reliable interaction with Anki via the local AnkiConnect HTTP API. It translates user requests into JSON actions, crafts and runs them with curl/jq, and enforces a strict confirmation policy for destructive operations. Value: safely automate deck, note, card, media, and sync management through a clear CLI workflow.
Renders Mermaid diagrams to SVG and PNG using the Beautiful Mermaid library and agent-browser. It validates/generates Mermaid code, renders vector SVG, captures high-res PNG via headless browser, and cleans up temp files. Outputs scalable SVG and 4K PNG for docs and sharing.
Retrieves up-to-date software library docs via Context7 API using curl. Searches library ID then fetches current documentation and code examples. Helps verify API usage and avoid outdated training knowledge for any framework.
Helps with GitHub Copilot SDK work across Node.js/TypeScript, Python, Go, .NET, and Java. Covers setup, auth, permissions, streaming, custom tools/agents, MCP, hooks, skills, and session persistence via official docs. Value: build Copilot-driven apps correctly with preview-aware, non-stale guidance.
A fast, script-friendly CLI skill for operating Google Workspace (Gmail, Calendar, Drive, Docs, Sheets, etc.) from the command line via gog/gogcli. Uses JSON-first output, multi-account auth, and agent-safe guardrails (dry-run, no-send). Enables automated, secure Google account/file/mail/calendar tasks with clear commands and references.
Generate and edit images via OpenAI GPT Image 1.5 using a Python script run with uv. Supports text-to-image, full edits, and mask inpainting with quality/size/background options. Outputs saved PNGs to the working directory for designers and creators.
Initialises a git repo in the current directory with optional agent commit instructions in AGENTS.md and a tailored .gitignore. Method: run git init, prompt user for commit rules and ignore patterns, then commit config files. Value: sets up version control quickly with guardrails for agent workflows.
Generates authentic lorem ipsum placeholder text via a bundled Python script. Method: invoke `uv run scripts/generate.py` with options for length (words, chars, tokens) and structure (paragraphs, headings, bullets, mixed, HTML). Value: produces consistent filler text for drafts, designs, or tests, saved to file or clipboard.
Converts PDF, Office, HTML, CSV, JSON, XML, images, audio, ZIP, YouTube and EPub files to Markdown via `uvx markitdown` with no install. Output preserves structure for LLM processing or text analysis, supporting file, stdout and stdin with optional Azure enhancement.
Semantic code search skill using mgrep for efficient exploration of large or nested codebases. It indexes repos via a watcher and supports natural language queries with options for content, AI answers, and sync. Value: complements grep by finding features and intent in unfamiliar code quickly.
Provides direct read-only access to the Monologue Notes REST API for listing, searching, and fetching the authenticated user's notes via curl/jq. Key methods: bearer-token auth, query filters, pagination. Value: secure, scriptable note retrieval without custom libraries.
Generate and edit images via Google's Nano Banana 2 (Gemini 3.1 Flash Image Preview) API. Uses a Python script run with uv to create images from text or edit with reference images, supporting resolution 512px–4K and explicit aspect ratios. Outputs PNG files in the project context for fast visual iteration.
Generate and edit images via Google's Nano Banana Pro (Gemini 3 Pro Image) API using a Python script. Supports text-to-image and image-to-image editing with 1K/2K/4K resolution. Outputs PNG files to the user's working directory, enabling fast AI image creation and modification from the command line.
This skill enables agents to interact with Notion workspaces via the Notion REST API using curl and jq. It covers auth, search, pages, blocks, databases, data sources, and users with rate-limit and confirmation handling. Value: programmatic, scriptable Notion automation and data ops without the UI.
Promptify transforms vague user requests into detailed, precise prompts optimized for AI models. It reads the input, plans needed context, and rewrites it as specification-style markdown following strict structure and language rules. Value: produces unambiguous, ready-to-use prompts that improve AI response quality.
This skill enables Agent to interact with Raindrop.io bookmark service via REST API using curl and jq. It covers auth, raindrops, collections, tags, highlights, filters, import and backups. Value: programmatic bookmark management with safe destructive-action confirmation.
Generates beautiful code snippet images via ray.so for sharing. Verifies agent-browser, fetches themes/languages, asks all styling params, builds a hash URL, and captures a 4x PNG with agent-browser. Outputs a high-res local image styled to user preference.
Provides a curl-based interface to Tavily's REST API for web search, extraction, mapping, crawling, and research. Uses endpoints like /search and /extract with API key auth to return LLM-friendly structured results. Value: enables live web data retrieval and multi-step research when no built-in tool is available.
This skill provides procedural guidance for interacting with Todoist via the `td` CLI. It covers CRUD operations for tasks, projects, sections, labels, comments, and more, with JSON output support and confirmation before destructive actions. Value: lets agents and users manage Todoist efficiently from the command line with safe, structured workflows.
A novelty skill that displays colorful ANSI terminal art of the word "ultrathink" when triggered by the user saying "ultrathink" or invoking /ultrathink. It runs a single printf Bash command with hardcoded escape sequences and produces no other output, serving as a lightweight visual flourish for agent interactions.
A CLI skill to read/write Upstash Redis-compatible KV store via REST API using a bun script. Provides Redis commands (strings, hashes, lists, sets, sorted sets, keys) with confirmation for destructive ops. Enables agents to persist state, cache data, and build queues or leaderboards.
Extract transcripts from YouTube videos via youtube-transcript-api using a Python script run by uv. Supports URL/ID input with or without timestamps, outputs verbatim text cleaned into paragraphs, saving as a txt file. Value: quick, reliable caption retrieval for analysis or notes.
