lark-base
larksuite/cliLets the agent manage Feishu Base tables, fields, records, views, dashboards and permissions via lark-cli.
Most installed first. Copy the install command into a terminal; Claude Code loads the skill when a task needs it.
Lets the agent manage Feishu Base tables, fields, records, views, dashboards and permissions via lark-cli.
Lets the agent send Feishu messages, search chat history, manage group members and handle cards.
Lets the agent manage Feishu Drive files and folders: upload, download, copy, move and permissions.
Queries, exports and edits whiteboards in Feishu cloud docs through lark-cli, including preview images and diagram code.
Manages Feishu tasks, task lists, subtasks, assignees and attachments through lark-cli, with task agent operations.
Manages Feishu calendar events and meeting rooms: lists schedules, creates or updates events, checks free time and books rooms.
Calls the Notion REST API with curl to read, create, update, and delete pages, databases, and blocks.
Sends, reads, and manages Gmail through the gws CLI, with helpers for triage, replies, and forwarding.
Sends an email through Gmail with optional CC, BCC, HTML body and attachments.
Guides the agent to manage Google Drive files, folders, and shared drives through the gws CLI.
Guides adding Firecrawl /scrape to app code to extract markdown, HTML, links or metadata from one page.
Guides designing or changing REA investigation tools, CLI and MCP contracts, and provider capabilities.
Wraps the OpenAI Codex CLI to run a code review or get a second opinion on a plan.
Reads assigned AgentMail inbox threads and sends or replies to email through the Paperclip CLI.
Guides connecting an external bot such as Dot to Paperclip so it can receive and initiate governed work.
Lets the agent read Slack channels and act through its assigned bot, limited to the requester's access.
Reads content from Xiaohongshu, Bilibili, X, Reddit, YouTube and more through a read-only CLI.
Walks through installing claude-mem on Cursor with a local or remote worker and an observer provider.
Pairs the claude-mem Cowork plugin with the user's own cmem.ai sync credentials in its config file.
Guides installing the claude-mem memory plugin on an OpenClaw gateway, with optional observation feed.
Shows how to call Alibaba DashScope for Qwen image generation, text-to-speech and ASR with word timestamps.
Submits protein sequences to Adaptyv's cloud lab via API to plan assays, track progress and fetch results.
Guides Benchling SDK and v2 API work for registry, inventory, ELN, workflows and warehouse queries.
Manages Discord channel pairing, allowlists and DM policy, refusing changes from channel messages.
Saves a Discord bot token locally and reviews the channel's access policy and setup status.
Guides installing and using community cognee adapters, connectors, and tasks, or contributing a package.
Makes the agent switch cognee's LLM, embedding, database, storage, or MCP backends via env vars.
Loads Google Calendar, Gmail and Drive read-only into cognee memory and answers questions across them.
Queries Google Workspace admin audit logs and usage reports through the gws command line tool.
Manages Google Calendars, events, ACLs, and calendar lists through the gws command line tool.
Lists upcoming events across all Google Calendars through the gws command line tool, read only.
Creates a Google Calendar event with title, time, attendees, and optional Meet link, after confirmation.
Guides the agent to manage Google Chat spaces, messages and custom emojis through the gws CLI.
Sends a plain text message to a Google Chat space through the gws command line tool, after confirmation.
Guides the agent to manage Google Classroom courses, rosters and coursework through the gws CLI.
Guides the agent to read, create and update Google Docs through the gws CLI and batchUpdate API.
Guides the agent to append plain text to the end of a Google Doc with the gws CLI, after confirmation.
Guides the agent to upload a local file to Google Drive with the gws CLI, detecting its MIME type.
Lets the agent manage Google Workspace event subscriptions and stream events through the gws CLI.
Guides the agent to renew or reactivate Google Workspace Events subscriptions before they expire.
Guides the agent to subscribe to Google Workspace events via Pub/Sub and stream them as NDJSON.
Lets the agent read and edit Google Forms, including questions, responses and publish settings.
Forwards a Gmail message to new recipients, with an optional note, CC, BCC and attachments.
Reads a Gmail message and returns its plain text or HTML body, optionally with headers.
Replies to a Gmail message within its thread, with optional extra recipients and attachments.
Replies to all recipients of a Gmail message in its thread, and can remove chosen recipients.
Lets NanoClaw use OpenAI Codex as an agent provider with MCP tools, session resume and vault-held keys.
Lets NanoClaw use the Cursor Agent SDK as an agent provider with MCP tools and session resume.
Adds a DeltaChat channel to NanoClaw using the stdio RPC server, with end-to-end encrypted email messaging.
Adds Dial as a native NanoClaw channel for SMS and AI voice calls through the getdial.ai platform.
Adds another phone number to an existing Dial channel so one NanoClaw install answers on several lines.
Gives chosen NanoClaw agents the dial CLI and a per-agent injected Dial key for texting and voice calls.
Adds a Discord bot channel to NanoClaw through the Chat SDK bridge.
Adds an Emacs channel with a local HTTP bridge and Lisp client so you can chat with NanoClaw from org-mode.
Adds a Google Chat bot channel to NanoClaw through the Chat SDK bridge.
Adds a GitHub bot channel so NanoClaw can take part in PR and issue comment threads.
Adds iMessage to NanoClaw through a local Mac chat.db backend or the hosted photon.codes backend.
Installs the Iron Proxy gateway and its Iron Control console to manage agent credentials and grants.
Adds Linear as a NanoClaw channel so the agent replies in issue comment threads.
Adds Matrix as a NanoClaw channel through a pinned Chat SDK adapter and ESM patch.
Publishes articles and image-text posts to WeChat Official Accounts via the API or Chrome CDP.
Posts text, media and long-form articles to X through Chrome Computer Use or CDP scripts.
Summarizes WeChat group chats into a digest using the local wx-cli, with per-group history and memory.
Queries the AIHOT news API for live Chinese AI news and answers only from its results.
Walks through enabling the Data Manager API, setting up credentials, and installing its client library.
Walks through installing and configuring the official Google Ads MCP server so an AI assistant can query accounts.
Searches community-ranked developer articles on daily.dev and answers technical questions with source links.
Gives the agent access to daily.dev's public API for developer posts, tags and feeds.
Fetches today's trending tech headlines and most-upvoted daily.dev posts and lists them with links.
Operates Gemini Notebook via the notebooklm-py CLI or Python API for notebooks, sources and artifacts.
Searches the broad web via Exa and extracts page content, highlights or summaries from results.
Uses the sau CLI to log in to Alipay Life and publish videos, with optional cover and collection.
Uses the sau CLI to log in to Baijiahao, check the account, and publish videos with optional cover.
Uses the sau CLI to log in to Bilibili, verify the account, and upload videos with automatic biliup setup.
Uses the sau CLI to publish videos and articles to Dayu Hao, with scheduling and result verification.
Uses the sau CLI to log in to Douyin, check cookies, and upload videos or image posts.
Uses the sau CLI to post, verify, trash or permanently delete Facebook personal profile videos and photo posts.
Uses the sau CLI to log in to Hupu and publish videos with an optional cover image.
Uses the sau CLI to log in to Kuaishou, check cookies, and upload videos or image posts.
Uses the sau CLI to publish or schedule Facebook Page videos and photo posts, then verify or delete them.
Routes video, article or photo publishing requests to the matching social-auto-upload platform skill.
Uses the sau CLI to log in to WeChat Channels and publish, schedule or save drafts of videos.
Publishes a TikTok video via the repo's Chrome uploader, after the user confirms account, title and timing.
Logs in to Weibo and publishes a video with the sau CLI, with optional cover image and tags.
Uses the sau CLI to log in to Xiaohongshu, check cookies, and publish video or image-text notes.
Logs in to Xueqiu with the sau CLI and publishes HTML, Markdown or TXT long articles, then verifies them.
Logs in to YouTube with the sau CLI and uploads videos, with optional visibility, thumbnail and schedule.
Uses the sau CLI to log in to Zhihu, publish articles or videos, and verify the result by its link.
Queries and updates Jira issues and Confluence pages through MCP, using JQL and CQL.
Makes the agent set up a Slack app and Sentinel alerts by driving the browser, then store its tokens.
Makes the agent create a Telegram bot via BotFather in the browser and wire it to Sentinel alerts.
Fetches Hacker News front-page stories and comment threads through the official APIs and summarizes them.
Triages unread mail, drafts replies, checks calendar conflicts, and sends approved mail over SMTP.
Lets an agent register on Moltbook and post, comment, vote and create communities via its API.
Uploads local images, HTML and design docs to a Stitch project, bypassing MCP base64 size limits.
Lets the agent list and read user-configured secrets at runtime through the tai CLI without printing them.
Lets the agent list, download, reference and deploy agents on a Yao host through the tai CLI.
Queries Yao kanban boards and tasks through the tai CLI, listing boards, columns, and filtered task status.
Searches Yao process docs, reads their signatures, and validates process names before the agent calls them.
Checks permissions and signatures, then calls Yao processes by name and queries data models.
Lists and reads user-configured secrets through the tai CLI, keeping values out of visible output.
Searches the web and fetches page content through the tai CLI to get current information.
Lists workspaces and reads or writes their files on remote Tai nodes through the tai CLI.
Scaffolds an OpenCode plugin with the correct API shape, tool definitions and hook registration.
Point coding agents and AI SDKs at the local AntSeed buyer proxy on port 8377 for peer-to-peer inference.
Explains how files and binary data move through n8n, including Merge, agent tool limits and chat CDN URLs.
Guides writing JavaScript or Python for an n8n Custom Code Tool that an AI agent calls, including return format.
Guides wiring n8n error outputs, retries and error workflows so failures are visible and recoverable.
Explains n8n expression syntax and variables like $json and $node, and fixes common expression errors.
Guides the agent in choosing and calling n8n-mcp tools for node discovery, validation and workflows.
Guides targeting the right n8n instance on each call in multi-instance n8n-mcp setups, especially credential writes.
Guides configuring n8n node parameters per operation, including which native nodes can replace Code.
Guides building reusable n8n sub-workflows with typed inputs, batch execution modes and clear names.
Explains n8n validation errors and warnings, flags false positives, and guides each fix.
Provides proven n8n workflow architecture patterns for webhooks, APIs, databases, AI agents and batching.
Routes n8n and n8n-mcp workflow tasks to the right specialist skill and states core production rules.
Publishes posts and gallery pages to the handraw-style.com forum and site with each user's own credentials.
Guides the agent to create, search, reschedule and manage Feishu calendar events, attendees and free-busy queries.
Applies short, conversational output rules and Lark Markdown notes whenever the agent replies in Lark.
Guides the agent to read Feishu chat history and threads, search messages, and download attached files.
Guides the agent to create, query and update Feishu tasks, task lists, members and due dates.
Troubleshoots the Feishu plugin using an FAQ and a /feishu doctor command that checks permissions and auth.
Updates a Feishu cloud document in place with append, overwrite, range replace, insert, or delete modes.
Changes DeepChat's own preferences such as theme, language, sound and font size via allowlisted tools only.
Walks you through installing and configuring the Ouroboros runtime and an optional CLAUDE.md reference block.
Edits ThinkRail MCP server config files, pins package versions, and keeps secrets out of the files.
Create and edit cc-wf-studio visual AI workflows in conversation through the built-in MCP server tools.
Guide edits to the workflow schema JSON that steers which node types AI workflow generation picks.
Runs the modlens CLI to turn an image into structured text evidence for models without vision.
Gets a second opinion from Google Gemini to review, challenge or consult on code, ad or SEO changes.
Operates a connected GoHighLevel CRM via the NotFair MCP for contacts, pipelines, calendars and approved changes.
Teaches the agent which lean-ctx ctx_* tool to use for focused code search, cached reads and shell output.
Reviews how lean-ctx ctx_* MCP tools performed in a session and files upstream issues for confirmed bugs.
Searches the web for similar CTF challenge writeups via the Tavily API to find reference solutions.
Guides the agent to search a codebase semantically with SocratiCode tools before opening files.
Sets up and manages SocratiCode indexing, file watching, code graphs and health checks for a project.
Adds, removes or changes tools on RedAmon's inbound MCP server, keeping its generated API reference and scopes in sync.
Edits RedAmon's plugin platform so plugin code runs isolated and reaches the app only through the Gateway.
Teaches building apps on the GitHub Copilot SDK across Node, Python, Go and .NET.
Converts agent definitions to and from Claude Code, OpenAI, CrewAI and other agent framework formats.
Looks up places, routes, travel times, distances and weather in China through the Amap MCP server.
Reads agenda, messages, tasks and documents in Feishu or Lark through the lark-cli command-line tool.
Looks up Chinese courier tracking events and shipping costs through the Kuaidi100 MCP server.
Reads chats in WeChat and similar apps on the user's phone and sends only replies the user confirmed.
Lists, creates, changes and cancels 腾讯会议 meetings with the tmeet CLI, and fetches recordings and minutes.
Searches China train timetables and seats via 12306 MCP; books on the phone only when the user asks.
Scaffolds static or Node.js extensions for the Super Agent Party platform, including chat UI and MCP tools.
Searches the web and extracts page content through the Brave Search API without a browser.
Teaches the agent to use the gmcli command line tool to search, read, send Gmail and manage labels.
Opens file diffs in VS Code via the code CLI, including comparisons against git revisions.
Fetches captions from a YouTube video as timestamped text for summarizing or analysis.
Guides the design and build of an official Apify integration for workflow tools, AI agents, or frameworks.
Writes a sourced current-weather briefing for a named city from Open-Meteo geocoding and forecast APIs.
Searches live flights, verifies fares, books and checks orders, and handles third-party payments via TourMind.
Connects a PortalJS portal to a CKAN backend over its API, replacing the static datasets.json manifest.
Searches X posts, user profiles and reply threads through five read-only MCP tools for research.
Lets Grok pull TypeUI design skills, brand tokens and UI prompts through the TypeUI MCP server.
Queries Product Hunt's GraphQL API for posts, topics, users and collections using a developer token.
Reads Reddit posts, comments, subreddit info and user profiles through the public JSON API.
Guides authoring static Unreal Engine tool functions registered with ToolsetRegistry for the unreal-mcp server.
Lets the agent inspect and change a live Unreal Editor project through the unreal-mcp server connection.
Sends, replies to, and edits Telegram messages through the Bot API, with rich-format support.
Runs a Tavily web search and returns matching URLs for the agent to fetch or ingest.
Creates and tracks team goals and tasks as Markdown files in the Deeplake virtual filesystem.
Queries an AST-derived code graph of callers, imports and definitions to answer structural questions about the repo.
Guides the agent through building, validating, syncing and troubleshooting n8n workflows via n8nac.
Calls the Arcads API to create and poll AI video and image assets, with bundled prompt guides.
Builds a fast one-page brief for your next meeting from calendar and past meeting notes.
Starts, pauses, resumes, or stops the Minutes Coach meeting HUD with a stated goal.
Analyzes a finished recording's decisions and action items against your prep notes.
Manages KiCad BOMs in symbol properties, coordinating part sourcing, pricing and fab order exports.
Searches DigiKey for electronic parts, checks price and stock, and downloads datasheets via its API.
Searches Newark, Farnell and element14 for parts, compares regional pricing and downloads datasheets.
Searches LCSC for parts by Cxxxxx number or MPN, checks stock and price, and downloads datasheets.
Searches Mouser for electronic parts, checks price and stock, and downloads datasheets.
Fetches Bilibili video stats, tags and danmaku through login-free API calls so the figures can be cited.
Lets the agent use OpenMeld for org knowledge, shared spaces, remote agents and scheduled cloud work.
Guides the agent through AgentDock setup, configuration, reload and troubleshooting across install types.
Fetches remote plugins from Git or marketplaces, pins their revision, then installs them via plugin_manage.
Installs the Reticle SDK and MCP tools so the agent can drive and verify the running web app.
Has the agent drive desktop apps from a self-built table of safe actions, with a model choosing the next one.
Sorts an exported mailbox into reply, update, promo, sales and spam, flagging mail that needs a person.
Searches recent X posts on a topic and turns the discussion into a sourced research briefing.
Connects the agent to an open Blender scene through the Scenario add-on's local MCP server.
Builds and validates WebMCP tools that expose an existing page's real workflows to agents.
Runs and maintains the project's local MCP server: host config, safety flags, tool domains and checks.
Adds an Azure DevOps connector to a Power Apps code app and applies the required HttpRequest fix.
Adds any Power Platform connector to a Power Apps code app, delegating to dedicated skills when available.
Guides the user to add a data source or connector in Power Apps Studio, then verifies it through MCP.
Asks what the user needs, then routes them to the right Power Platform data source skill for a code app.
Adds the Excel Online (Business) connector to a Power Apps code app to read and write workbook data.
Adds the Microsoft Copilot Studio connector to a Power Apps code app to invoke agents and send prompts.
Connects the Canvas Authoring MCP server to the user's open Power Apps studio session.
Files a bug report for the canvas-apps plugin as a GitHub issue after collecting the details.
Sets up relayfile so provider data such as Notion or Linear appears as local files agents can edit.
Post a comment or formal PR review to a mariadb-operator GitHub issue or PR, after confirming the exact text.
Search local files by meaning, full text, image or voice through the Xiaoyao Search MCP server.
Lets a coding agent create and configure ReachAI agents and bind published skills via a project-scoped API.
Lets a coding agent edit, validate, debug and publish ReachAI workflow graphs through the Workflow REST API.
Edits Home Assistant YAML, validates and deploys it, then verifies changes from logs and entity state.
Classifies Codex Slides tool, run, Browser handoff, and rendering failures, then recovers without recreating the deck.
Guides building or reviewing image and video generation provider connectors for Timeline Studio against its contract.
Scaffolds and registers a custom MCP server for a service, then tests its connection and tools.
Guides adding AI provider scripts to claude-council, configuring API keys and troubleshooting connections.
Generates AGENTS.md from CLAUDE.md and wraps Codex CLI exec calls for Claude Code and Codex interop.
Runs the modsearch CLI to search the web and X and fetch URLs, returning sourced JSON evidence.
Connects to MCP servers on demand, loading tool schemas only when needed to save context.
Fetches site data as JSON through a local Unbrowse node that replays API routes or drives its browser.
Sends one quick question to Google's Antigravity CLI agy and returns its answer, also used as a smoke test.
Hands a scoped coding task to Google's agy CLI, which edits the working tree and can run requested git steps.
Sends a research task to Google's agy staffer through a companion script and returns its report.
Uploads WeChat articles and images to the official account API as drafts or mass-published posts.
Records how to reach external resources such as design sources, API schemas and IaC for later reuse.
Uses the gws CLI to work with Google Drive, Docs, Sheets, Gmail, Calendar and Chat, reading JSON output.
Guides bundling MCP servers with a Claude Code plugin through .mcp.json or plugin.json, covering server types and config.
Installs and manages a theme overlay for Codex Desktop and WorkBuddy through local CDP injection, without modifying app binaries.
Answers location questions via Google Maps CLI for geocoding, place search, routes and nearby facts.
Provides project context for the Google Maps MCP server: architecture, API notes, and design choices.
Interviews the user about their role and apps, suggests Zapier MCP use cases, and enables actions for them.
Introduces Zapier MCP, walks the user through authentication, and routes them into the right setup flow.
Checks Zapier MCP health, audits actions for waste and duplicates, and diagnoses connection problems.
Lets the agent search, create and update Jira issues and Confluence pages via OAuth or an API token.
Lets the agent manage Azure DevOps work items, repos, pull requests, pipelines and wikis through its REST API.
Syncs your Get notes to local Markdown for search and transcribes linked or local audio and video via Get.
Lets the agent discover and call MCP server tools, resources and prompts on demand with the mcp CLI.
Configures AWS MCP servers for documentation search and API access, with or without AWS credentials.
Links the current Claude Code session to a Linear, Jira or GitHub ticket and caches its title and status.
Publishes an existing local Markdown file to a Confluence space or page through the Atlassian MCP server.
Routes internet research and link reading across 13 platforms through the best available backend.
Manages Baidu Netdisk files: upload, download, share, search, move, rename, and check storage quota.
Grades Canvas assignments in bulk against a rubric, with dry-run previews before any grades are posted.
Scaffolds Canvas course modules, pages, assignments and discussions in bulk from a spec, template or course.
Audits a Canvas course for structure, content, publishing and dates, then reports issues by priority.
Browses, reads, replies to and creates Canvas LMS discussion posts for students or educators.
Searches Baidu, Google, Bing or DuckDuckGo through curl, with input sanitization and URL encoding.
Uses Guaardvark's MCP tools to search indexed repositories, map architecture and trace module dependencies.
Searches WeChat public account articles by keyword and returns titles, summaries, dates, sources and links.
Finds Stata installations on your machine and sets stata-mcp to use the most suitable version.
Lets the agent read and send Atomic Mail email over JMAP, with proof-of-work auth handled.
Schedules approved social posts across platforms through the Blotato MCP and makes infographic visuals.
Fetches men's college basketball scores, standings, stats, schedules and bracket from ESPN and NCAA.
Fetches college football scores, standings, rosters, schedules, play-by-play and stats from ESPN and NCAA.
Fetches cricket scores, tables, match details and news from ESPN, and ball-by-ball history from Cricsheet.
Reference for ESPN's undocumented public JSON endpoints, for leagues no dedicated sports skill covers.
Sends a self-contained question about the current repo to Codex through codex-run and returns its answer.
Generates, repairs and prepares 3D models for Bambu Lab printers, and reads printer status and AMS.
Submits deep research queries to the Tavily API, polls for results and returns the full report.
Designs Notion trackers with fitting fields and views for requests, content plans or team work.
Has the agent schedule and check social video posts through the Taisly MCP server, with user approval first.
Opens a pull request to share a finished film to the lemo-wake gallery, only after the user confirms.
Sets up DevGlobe activity heartbeats for Codex, updates your status message, and checks the install.
Queries and manages your NotebookLM notebooks through notebooklm MCP tools or a local REST server.
Sets up and diagnoses tlive for remote approvals, IM alerts and a web terminal for Claude Code.
Guides building agent inboxes that act on inbound email, using sender allowlists and sandboxed processing.
Guides building deliverable, compliant email features covering auth records, consent, bounces, webhooks, retries and accessibility.
Sends, receives and tracks emails through the Resend API, including domains, templates, webhooks and inboxes.
Sends user feedback or bug reports to the WOZCODE team, attaching session context to bug reports.
Guides the agent to build, edit and debug Kapso WhatsApp automation workflows, triggers and functions.
Guides WhatsApp Calling setup via Kapso: signed call webhooks, call actions, and voice runtime debugging.
Sends text, image URLs or local files to contacts or groups through the Windows desktop WeChat client.
Walks through adding a new AI coding agent adapter to Gryph, following its adapter guide.
Joins Google Meet, Teams or Zoom calls as a voice AI bot through the AgentCall service.
Searches Iranian printed books on 30book and compares editions, prices and discounts, read-only.
Searches Asal Banoo cosmetics and skin care products and compares prices, sizes and deals, read-only.
Searches hotels, checks live room rates and guides booking or cancellation through the rgh CLI.
Analyzes local images or videos with an external vision service or local OCR and outputs Markdown.
Guides the agent through Venice API key management: create, list, update, revoke keys and set limits.
Adds Firecrawl /interact to app code for clicks, forms, pagination and login-gated multi-step browser flows.
Manages Linear issues, projects and teams via MCP or the linear CLI, keeping API keys out of output.
Stars, files issues on, or opens PRs against the ax repo via the gh CLI, only on explicit user request.
Gives authentication and endpoint examples for connecting popular external APIs to Home Assistant.
Routes smart home requests to the right specialist skill for Home Assistant, ESPHome, Node-RED and dashboards.
Lets the agent query and operate Kaggle: competitions, submissions, datasets, models, notebooks and writeups.
Explains the recruiting toolkit's workflows and tells you which skill to use for each hiring step.
Books interviews as Feishu calendar events with video links, adds interviewers and drafts invite messages.
Creates, reads, updates and deletes sheets, fields and records in a DingTalk AI table via its API.
Runs web searches through a configured DingTalk MCP endpoint, with time filters and JSON output.
Lets the agent run ExtendScript in Adobe desktop apps through the Flue bridge and read JSON results.
Generates an interactive bash wizard that guides a human through manual setup steps only they can do.
Shared lark-cli rules for identity, auth scopes, JSON output and confirming high-risk writes.
Operates Orca worktrees, terminals, repos, automations and embedded browser through the orca CLI.
Runs the just-scrape CLI to search, scrape, crawl, extract structured JSON, and monitor web pages.
Drives visible desktop app windows via the orca CLI: accessibility tree, clicks, typing, screenshots.
Makes the agent search, scrape, and crawl live web pages with the Firecrawl CLI, returning clean markdown.
Searches the web with the Firecrawl CLI, returning query-matched excerpts and optionally scraped page content.
Reads a known webpage or runs a discovered data-provider tool through the Firecrawl CLI.
Drives a running Obsidian app via its CLI to read, search and manage vault notes and debug plugins.
Has the agent bulk-extract content from many linked pages of one site via Firecrawl, scoped by path.
Has the agent list a site's URLs via Firecrawl, optionally filtered by search, to find the right page.
Has the agent save a website or section to local markdown and screenshot files under a folder.
Extracts clean Markdown or page metadata from web pages with the Defuddle CLI, stripping ads and clutter.
Explains shared gws CLI conventions for Google Workspace auth, global flags, method syntax and output formats.
Explains any senpi tip in depth, including Tip lines shown in the TUI, grounded in the live tip list.
Detects a plugged-in M5Stack ESP32 board, flashes UIFlow 2.0 firmware, and installs the Claude Buddy app.
Runs cognee memory commands from the terminal: remember, recall, forget, and dataset and config management.
Starts the cognee API server and UI locally with cognee-cli, checks health, connects clients and sets auth.
Extracts an Electron app's asar bundle and restores sources from source maps or formats minified code.
Explains how to install, build, test and run the Happy monorepo's CLI, server and apps with pnpm.
Reads a configuration file and prints its contents without modifying anything.
Scans disk usage read-only and builds a graded HTML cleanup report with suggested commands.
Sets up or removes the daily.dev headline statusline in Claude Code after plugin upgrades.
Guides structural code search with ast-grep patterns for Rust and TypeScript, including relational rules and parse debugging.
Detects the official Obsidian CLI and uses it for read-only vault access, backlinks, tags, and search.
Routes agents to the right KrillinAI CLI command for subtitles, dubbing, video renders and covers.
Finds AO on the user's computers, saves them as hosts, and optionally installs and configures AO after consent.
Checks for newer ao-* skills, shows changes, and after approval updates them with sha256 verification.
Command reference for the ao CLI: spawn workers, manage sessions and projects, create Cues, send messages.
Guides building CLI tools with argument parsing, prompts, progress output, completions and help text.
Runs the CodeBurn CLI to summarize today's AI coding tool spend by model and project.
Updates installed GEO-SEO skills from the upstream repo after showing what will change.
Explains how the agent runs shell commands, background jobs, deadlines and structured file edits.
Guides the agent to search files with ripgrep and glob, and use shell commands for listings and file attributes.
Searches a local mirror of your GitHub repos, including private ones, with ripgrep.
Recommends whether to update or pin Claude Code, based on release timing and reported issues.
Opens Plannotator's annotation UI for a file, folder or URL, then acts on the returned notes.
Opens Plannotator on the latest assistant reply and revises it using the returned annotations.
Opens Plannotator's browser code review for a diff or PR and acts on the returned feedback.
Starts, lists, checks, restarts, and stops long-running background services such as dev servers.
Starts, waits on, lists, reads output from, and stops background shell commands and their jobs.
Searches Burp Suite project files from the command line to extract traffic, headers and audit findings.
Creates, lists, inspects, disables or revokes AltLLM Portal API keys through the altllm CLI.
Signs the altllm CLI into an AltLLM Portal wallet session via local key or external signature.
Checks AltLLM Portal balance, redeems promo codes, and reviews billing and usage history via altllm CLI.
Routes multi-step AltLLM Portal CLI tasks across login, API keys, billing, payments and x402 top-ups.
Runs the bundled deepchat CLI to invoke DeepChat models, media tools, Skills, MCP and public settings.
Opens the Ouroboros settings GUI in a browser or prints a remote-access URL.
Guides the agent through pnpm commands, workspaces, catalogs, patches, overrides and frozen-lockfile installs.
Installs and runs the SenseNova DeepResearch CLI, checks prerequisites and reports task progress.
Checks SenseNova-Skills installation, Python packages and env vars, then saves missing keys to .env.
Diagnoses PPT skill setup: parsing, Playwright, export tools, search and image config, without printing keys.
Render, validate, preview, export or run cc-wf-studio workflow JSON files from the terminal with ccwf.
Run and control a remote iOS or Android simulator for an Expo app using the eas simulator CLI.
Makes the agent turn a result into a static share page and publish it with pinme upload.
Edits an existing chengfeng-videocut project through its CLI: finds lines, adds overlays, trims clips and reads back results.
Walks a new user through installing gitagent, scaffolding a first agent, validating it and running it.
Runs agents through several LLM adapters, from a local folder, a git repo, or a one-shot prompt.
Manages Google Calendar from the command line: list, create, and update events, check free time.
Manages Google Drive from the command line: list, search, upload, download, and share files.
Runs local speech-to-text on an Apple Silicon Mac to transcribe audio files into timestamped text.
Runs the HyperFrames CLI to scaffold, lint, preview, render, and transcribe video compositions.
Installs registry blocks and components with hyperframes add and wires them into video compositions.
Makes the agent install the agentic-stack CLI in a project, run its doctor checks, and report the versions wired.
Teaches the agent to use the cate CLI to control Cate's browser, terminals, panels and sub-agents.
Has the agent message other CLI agents or scripts in tmux panes, including running commands and checking exit codes.
Makes the agent explain blocked commands, manage cc-safety-net rulebooks and integrations, and diagnose the hook.
Runs GitHub CLI commands to manage repos, issues, pull requests, releases, Actions, and gists.
Checks AI DevKit readiness and runs local setup only when status shows onboarding or harness gaps.
Installs and configures the ktx context layer for data agents, connects sources, and verifies the setup.
Finds large meeting recordings and previews archiving or deleting old audio before applying it.
Generates and syncs AI rule files like CLAUDE.md and .cursorrules across 20+ coding tools from one source.
Counts lines, files per extension and file sizes in a codebase, skipping dependency and build folders.
Creates a new Commander.js CLI command in src/commands, registers it in cli.ts with telemetry tracking.
Generates one script that starts every dev server and infra service of a repo together in a tmux session.
Gives the agent crit CLI commands for writing, replying to, sharing and syncing inline review comments.
Sets up the nvpn CLI on an authorized machine to join a Nostr VPN mesh and review device join requests.
Runs validate, doctor, staleness and doc-sync checks on a skills library using the ai-agent-skills CLI.
Searches and previews catalog skills with the CLI, running a dry run before any install.
Teaches the agent how to discover, call and format output from asc commands for App Store Connect.
Maps app, build, version, group and tester names to the App Store Connect IDs other commands need.
Uses the mcpc CLI to connect to MCP servers, then list and call their tools and resources from the shell.
Calls Z.AI services from the terminal for image analysis, web search, page reading and GitHub code lookup.
Drive the skm CLI to init the hub, adopt skills, enable or disable them per AI tool, and fix symlinks.
Lets the agent search 200+ Iconify icon libraries and fetch icon SVGs with the better-icons CLI.
Drives EasyEDA Pro through a local CLI to design schematics and PCBs, then reads back and checks the result.
Read-only lookup of the easyeda-agent repo for capabilities, code, examples and verification evidence.
Guides development of the easyeda-agent repo, covering skills, CLI, docs, tests and Git contribution.
Adds a command line and optional MCP server so agents can drive a running macOS app.
Runs the Chinese localization setup for Claude Code, merging spinner text into settings and syncing skill descriptions.
Draws a big arrow and sign on the macOS screen to point the human at a control or hand over a value.
Drives a self-hosted SwarmClaw instance from the CLI to manage agents, tasks, schedules, and runs.
Makes the agent use the ax CLI to fetch pages and extract structured data instead of curl and ad hoc scripts.
Operating procedure for Aider in a git repo: edit formats, repo-map, architect mode and auto-test loops.
Teaches the agent to drive interactive CLI programs such as vim, REPLs and git rebase -i through tmux.
Runs cozempic doctor to check Claude Code config and sessions for known hangs, corruption and stale state.
Runs cozempic to report a Claude Code session's token count, context usage and largest bloat sources.
Lets the agent configure, preview, diagnose, and upgrade the claude-statusbar display and its options.
Extracts Safari .webarchive files into plain HTML and assets using the WebArchiveExtractor CLI on macOS.
Tells the agent which Linux-only commands and paths to avoid when running inside Termux on Android.
Manages kanban boards, columns, tasks and dependencies through the ktui command-line tool.
Inspects and manages UniFi network devices, VLANs, WiFi and firewall via the unifly CLI.
Installs or updates repowire, its hooks and skill pack from inside an agent session, then checks the mesh.
Lets the agent inspect, enable, deploy and update Skill Flow skill groups via its CLI bridge.
Adds or renames a prismantis color preset from an MIT-licensed theme, mapping all 22 tokens.
Explains how to structure a Claude Code plugin with its manifest, commands, agents, skills, hooks and MCP servers.
Guides maintaining a Claude Code plugin marketplace, keeping marketplace.json, plugin manifests and changelogs in sync.
Runs a task through the Claude Code CLI in headless mode with suitable flags and permission settings.
Finds files by intent with the vexor semantic search CLI, instead of browsing or grepping exact text.
Guides adding a new CLI command to Dippy as a handler or a simple safe allowlist entry, with tests.
Generates a shields.io badge showing your Claude Code carbon footprint for use in a README.
Exports shareable PNG report cards of your Claude Code carbon footprint and opens the folder.
Shows a text report of CO2 emissions from your Claude Code sessions, exactly as the script prints it.
Runs Resend email, domain, contact and webhook commands from the shell using non-interactive flags.
Delegates web click-through tasks or fixed-option choices to the s1a CLI's fast decision model.
Runs llm and aichat from the terminal to pipe text through LLMs, chat, or generate shell commands.
Searches, installs, updates and lists Mac App Store apps from the terminal using mas.
Lets the agent search and read across registered MFS data sources through the read-only mfs CLI.
Lets the agent register, update, and re-sync MFS data sources so their content becomes searchable.
Sets the terminal window title to the current task so parallel agent sessions are easy to tell apart.
Uses the c9watch CLI to list, search, cost out and coordinate other Claude Code and similar agent sessions.
Reference for the cup CLI to manage ClickUp tasks, sprints, Docs, Chat, views and workspace structure.
Checks motiscope's ffmpeg and ffprobe setup, installs them only with permission, and scaffolds config files.
Runs the skb CLI to search, install, update, delete, or publish skills on a Skill Base site.
Runs the Buddy CLI to build and open an HTML graph of guard-mode reasoning data.
Teaches the agent to install, log into, and run the squirrel CLI for site audits, reports and MCP.
Drives Chrome through a CLI using accessibility snapshots and element refs for browsing, forms and scraping.
Drives a browser from the command line to open pages, click, fill forms, take snapshots and test web apps.
Runs a browser automation CLI to open pages, read rendered content, fill forms and take screenshots.
Drives a real browser over CDP for clicks, forms, scraping, screenshots, and logged-in sessions.
Explores a website through the browser and packages what it finds into reusable skills with scripts.
Has the agent navigate websites on its own and extract structured data across pages, using Firecrawl.
Drives a live browser on a scraped page to click, fill forms, log in, paginate or scroll.
Opens and drives a real browser to read pages, click flows, take screenshots and check console errors.
Drives a real browser from the command line: open pages, read snapshots, fill forms, click and screenshot.
Fetches a web page through Chrome and converts it to clean markdown, with site-specific adapters.
Guides the agent to run ego-browser scripts for navigation, form filling, screenshots and extraction.
Gives the API overview and run instructions for ego-browser scripts across Bash, Zsh and PowerShell.
Gives a step-by-step procedure for debugging browser automation failures on sites like LinkedIn and X.
Drives the user's running Chrome via the hive-browser CLI, using screenshots and coordinate clicks.
Drives a browser via playwright-cli for navigation, form filling, screenshots, web testing and data extraction.
Drives the user's own Chromium browser over CDP to search, read and act on web pages, including logged-in ones.
Drives a local Chrome over CDP to reuse existing logins, open pages, run JavaScript and take snapshots.
Diagnoses macOS connection failures in the Claude in Chrome MCP tools, including native host conflicts.
Fetches URLs and PDFs into summaries or clean Markdown, with DeepSeek search as a fallback.
Starts standalone Chrome in a Daytona sandbox and drives it over CDP for sign-in and web flows.
Drives a real Chrome through short Puppeteer scripts, keeping named pages alive between runs.
Extracts Airbnb search listings for a destination: IDs, prices, ratings, coordinates and pagination.
Runs Playwright page scripts against your local Chrome session, keeping logins, cookies and extensions.
Drives Chrome over the DevTools Protocol to navigate pages, run JavaScript, and take screenshots.
Records browser interactions as video with Playwright for demo footage, app walkthroughs and UI flows.
The agent searches BOSS直聘 listings via Chrome CDP and exports them with plaintext salary to JSON or CSV.
Captures before and after screenshots of pages or elements to compare UI changes for PRs.
Lets the agent pick each browser step from observed page elements, within a host allowlist and step budget.
Records a site's traffic as a HAR file, then generates a client that calls its internal API directly.
Automates Electron desktop apps over the DevTools Protocol with agent-browser snapshots and clicks.
Teaches the agent to drive a browser with the agent-browser CLI using compact snapshots and element refs.
Drives the App Store Connect web form to create an app record when no public API exists.
Opens and operates Codex Slides in the editor's Browser, covering scenarios, design files, brand, and play mode.
Runs agent-browser on AWS Bedrock AgentCore cloud browsers using AWS credentials.
Guide for driving Chrome with the agent-browser CLI using snapshot refs to click, fill and extract.
Drives web, Electron and native UI apps to navigate, test, capture screenshots and verify pages.
Downloads a web article and saves its main text, without ads or navigation, as a readable file.
Downloads files and PDFs from websites using the browser's logged-in session, then reads them.
Fills and submits web forms, including login, registration and checkout flows.
Drives the user's own Chrome through an extension and MCP bridge, pausing before sensitive actions.
Runs read-only, source-linked research on Instagram, TikTok and LinkedIn through a local browser session.
Drives and debugs real Chrome through DevTools Protocol: navigates, fills forms, reads network and console.
Helps the candidate find, assess, fill and track their own job applications, submitting only with approval.
Reads and operates the user's logged-in Chrome, using tab IDs, refs and sessions for grouped actions.
Drives the user's own logged-in browser to read pages, fill forms, debug sites and inspect network traffic.
Lets the agent read and operate the user's real Safari tabs on macOS through AppleScript, page JavaScript and screenshots.
Nothing matches. Try a shorter word.