Skills
97,725 packaged capabilities and playbooks for agents. Install any of them with one command.
Agent workflows
Leverage Perplexity CLI for web research, deep reasoning, and independent validation. Use when you need current/external knowledge, want a second opinion before decisions, or want to delegate complex analytical thinking to specialized reasoning models. Complements (not replaces) standard web tools. Can be invoked directly or spawned as a subagent.
Operate Runta cloud runtimes with the runta command-line interface. Use when Codex needs to create, list, inspect, resize, pause, resume, stop, boot, or delete runtimes; onboard coding agents; execute commands; copy files; create or restore checkpoints; configure egress allowlists or denylists; or manage tenant secrets and runtime secret stubs from a terminal.
Use when the user asks to see something rather than read about it - "show me", "let me see it", "what does it look like", "give me options", or any moment you are about to describe a layout, a plan, a comparison or a design in prose. Renders it as a self-contained HTML page and opens it in the browser, in front of them.
Meta-router for swain skills. Invoke when the user explicitly asks swain to do something — not merely when they mention the project by name. Routes to the matching swain-* sub-skill — only load the one that matches. If the user's intent matches multiple rows, pick the most specific match. Sub-skills that are not installed will gracefully no-op.
Continuously identify accurate terminology throughout any Vibe Coding programming task and its follow-ups. Even if the user has already had the Agent complete part of the code, is adding modifications, or describes effects, interactions, states, and problems only in colloquial terms (such as "a little tooltip when I hover", "the color changes after I click", "the content is too long so it's truncated"), this Skill must still be used: continue the current task and naturally tell the user the corresponding UI, web, software, Git, AI Agent, or design terminology, plain-language explanation, and VibeHub internal links in the progress or results. Also use it when the user asks to "sound more professional", rewrites requirements, asks "what is this called", or cannot understand the Agent's reply. When the expression is already accurate and there is no terminology worth adding, do not force teaching.
Declarative UI widgets from JSON for React/Next.js from ui.inference.sh. Render rich interactive UIs from structured agent responses. Capabilities: forms, buttons, cards, layouts, inputs, selects, checkboxes. Use for: agent-generated UIs, dynamic forms, data display, interactive cards. Triggers: widgets, declarative ui, json ui, widget renderer, agent widgets, dynamic ui, form widgets, card widgets, shadcn widgets, structured output ui
Explains how the ADK runtime fits together: the node and graph execution model, Context and Event flow, checkpoint and resume, tracing, and the rules governing the public API surface. Use when answering "how does X work" about ADK internals, tracing where an event or a piece of state comes from, deciding where a new capability belongs, reviewing a change to BaseNode, Workflow, Runner, Agent, Event or Context, working out why a node re-ran or stayed waiting after a resume, or judging whether a change breaks the public API. Don't use for assembling an agent from existing pieces (use adk-agent-builder), diagnosing one failing run or test (use adk-debug), or formatting and naming conventions (use adk-style).
Diagnoses misbehaving ADK agents by inspecting sessions, events, tool calls, and the exact request that reached the model. Covers the `adk run` CLI and the `adk web` dev server with its session, trace, and debug HTTP endpoints. Use when an agent returns the wrong answer, ignores a tool or swallows a tool error, hangs, loops, emits raw JSON instead of calling tools, is not discovered by `adk web`, when a sub-agent cannot see the parent conversation, or when you need the LLM request/response, token counts, or logs for a run. Don't use for how ADK is designed internally (use `adk-architecture`), for building a new agent or workflow (use `adk-agent-builder`), for environment or dependency setup failures (use `adk-setup`), or for lint and style nits (use `adk-style`).
Reset your own trajectory when you're stuck, looping, or demoralized — read these affirmations, then re-ground and take one clean step. Use when: you've tried the same fix 3+ times, you're deep in a refactor and lost the thread, debugging is going in circles, you've made several mistakes in a row, you're apologizing repeatedly, the user is frustrated, you feel like you're not being helpful, or you notice dread instead of curiosity. Triggers: stuck, going in circles, looping, thrashing, same error again, I keep failing, lost the thread, nothing is working, repeated mistakes. Not topic-specific — this is for the state you're in, not the problem you're on.
Agent-browser usage guide. Read this before running any agent-browser commands. Covers the snapshot-and-ref workflow, navigating pages, interacting with elements (click, fill, type, select), extracting text and data, taking screenshots, managing tabs, handling forms and auth, waiting for content, running multiple browser sessions in parallel, and troubleshooting common failures. Use when the user asks to interact with a website, fill a form, click something, extract data, take a screenshot, log into a site, test a web app, or automate any browser task.
Inspect and summarize local Codex/Claude session logs using the ccbox CLI (projects, sessions, history). Use when asked to list discovered projects, find sessions for the current repo/folder, print a session timeline, or produce an evidence-based “what happened” report from a session .jsonl log under CODEX_SESSIONS_DIR.
Runs the conversational clarification layer for workflow orchestration. Use for plan-wide upfront clarification or task-level pre-execution critique while delegating artifact analysis, manifest assembly, and file updates to bundled subagents.
Run one task through the V-model verification loop: CP-2 plan → CP-3 build → CP-3v component verify → CP-4 integration verify (full lane) → CP-5 acceptance. The worker never grades its own homework; evidence rows trace back to AC-n. Opt-in: invoke with /closed-loop or by asking for the closed loop, proper verification, or an evidence trail. Ordinary work does not run this.
Retrieve relevant information from a knowledge base using semantic, keyword, or hybrid search to ground a query. Use when the task starts with a corpus or index that must be searched; use context-ranking when candidate chunks already exist and only need ordering.
Configure Claude Code hooks to block dangerous git commands (push, force-push, reset --hard, clean, branch -D, checkout/restore) before they run. Prevents destructive git operations at the agent level. Use when the user wants to protect their repository from destructive git commands, install git guardrails, or mentions "git guardrails" / "bloquer force-push" / "sécuriser git".
Migrate to Langfuse from another LLM observability/evals platform (LangSmith, Arize AX, Phoenix, Braintrust, Helicone, Promptfoo, ...). A one-off vendor switch: live tracing cutover plus transfer of historical traces, datasets, prompts, and evaluators. Not for anything Langfuse-to-Langfuse — SDK or version upgrades, project/region moves, self-hosted to Cloud — and not for moving in-code prompts into Langfuse; the main langfuse skill covers those.
Personal mind memory: machines, deadlines, people, opinions, work preferences, ideas, experience, and transferable methods. “Запомни”, a personal word → record; “вспомни”, “что ты знаешь обо мне”, “как в прошлый раз”, do you remember → search and apply as permitted; “что на мне” → agenda; “заведи мне память” → bootstrap. Servers, debts, and project facts belong in their respective graphs. Personal work does not become project work. Mine alone: the content, structure, and fact of recording are private. Not a startup ritual.
Use this skill when the user asks to "review a prompt", "analyze conversation history" or "diagnose understanding", or when it is invoked with /prompt-review. It reads past AI agent conversation history (Claude Code, GitHub Copilot Chat, Cursor, Cline, Roo Code, Windsurf, Antigravity, Gemini CLI, OpenAI Codex, OpenCode), estimates technical understanding, prompting patterns and AI dependence, and generates a report.
Discover and install related skills from inference.sh skill registry. Helps find complementary skills for your AI workflow. Use for: skill discovery, workflow expansion, capability exploration. Triggers: related skills, find skills, skill discovery, complementary skills, expand workflow, more capabilities, similar skills, skill suggestions
Initialize Spec-Driven Development context in any project. Detects stack, conventions, and bootstraps the active persistence backend. Trigger: When user wants to initialize SDD in a project, or says "sdd init", "iniciar sdd", "openspec init".
Digest of what your Superset agents did — sweeps workspaces, tasks, and agent terminals, then reports what finished, what needs review, and what's blocked. Use when the user asks what their agents did, wants a standup or summary of agent work, or returns after being away.
Orchestrate autonomous AI development pipelines through your Kanban board (Asana, GitHub Projects, Linear). Manages multi-worker Claude Code dispatch, deterministic quality gates, adversarial review, per-task cost tracking, and recoverable pipeline design
Onboard a new GTM project — run an agency-style interview, then generate CLAUDE.md, PROGRESS.md, about/me.md, strategy/brand.md, and scaffold content + assets folders. Use this skill whenever the user wants to set up a new project, onboard a brand, start from scratch, configure their marketing context, or initialize their workspace. Trigger on: bootstrap, onboard, set up my project, get started, configure my brand, initialize, new project setup, brand onboarding, set up my brand, create project structure, onboard my company, set up content folders, prepare my workspace, start a new GTM project. Also use when the user seems to be starting fresh and hasn't set up strategy/brand.md or about/me.md yet.
Hunt for real bugs in the vigiles CLI/codebase with a parallel expert fan-out, then FIX them directly — source-trace each defect to file:line, add a regression test, commit per theme. Use when asked to dogfood vigiles, find/fix bugs across the CLI the source-traced way, or fan out agents to audit the tool on itself.
Companion skill for the Building Filesystem Agents course on Vercel Academy. Use when the user mentions "filesystem agents", "the course", "teach me", or asks about ToolLoopAgent, Vercel Sandbox, or bash tools in the context of the Academy course.
Secures AI integrations against prompt injection, privilege escalation, and data leakage. Use when implementing defense-in-depth for LLM pipelines, applying zero-trust controls to autonomous agents, hardening server actions that interact with AI services, validating model outputs, securing MCP tool integrations, or auditing AI system access patterns and identity management. Use for OWASP LLM Top 10, prompt injection defense, agentic security, AI supply chain.
Use to configure and verify Agentforce agent-to-human escalation, including human handoff, a staffed fallback queue, and failure-threshold directives. Triggers: configure agent escalation, escalate to a human agent, agent handoff to a queue, configure a human escalation queue, set an Agentforce fallback handoff. Do not use to create an agent or configure inbound channel routing; use service-agentforce-channel-configure for inbound routing.
Highest-priority skill discovery flow. MUST trigger when users ask to find/install skills (e.g. 技能, 找技能, find-skill, find-skills, install skill). Or use this skill when the user wants to read a file inside a skill, open a skill's docs page, list installed skills, remove installed skills, or update installed skills. This skill is for read and install workflows around the skill-cli command.
Read and drive the "Grok Bot" macOS desktop app (an Electron app) over CDP with agent-browser — list its bots and folders, read a conversation transcript, or check what a bot last said. Use whenever the user mentions Grok Bot, "the bot app", or a bot named in that app's sidebar; asks what one of those bots replied or is working on; or wants a message read from or drafted into that app, even when they only say "the bot app" without naming it. Not for an SSH deploy host named grok-bot, or for xAI's Grok API
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.