AI Agents

Package this agent as a shareable Claw for someone else to review

Create, repair and validate SKILL.md agent skills and their bundled resources

Learn how to find and invoke skills before starting any conversation

Create, edit and test agent skills to confirm they work before deploying

Find out which skill in this collection fits your task and get routed to it

Distill principles repeated across skills into shared rule files

Measure whether agents follow their skills and rules, with compliance rates

Diagnose a draft prompt and rewrite it into a ready-to-paste version with reasons

Write model-facing tool definitions and skill context so models find and use information well

Evolve prompts, regexes, SQL and code with an evolutionary search loop

Create a reusable VS Code prompt file for a task you repeat often

Prompt a critical check of drafted answers, estimates and claims before use

Search and retrieve AI prompt templates from the prompts.chat library

Summarize a long voice ramble back for approval before acting on any of it

Delegate coding tasks to Codex, Claude Code or OpenCode as background workers

Dispatch two or more independent tasks to run in parallel agents

Run implementation plans as independent tasks handed to subagents

Pursue a long goal in one session by coordinating several subagents

Plan agent-run engineering with eval-first checks and cost-aware model routing

Build a persistent multi-agent setup with file memory, commands and schedules

Run Claude Code as an autonomous agent with memory, schedules and a task queue

Hand the current conversation to a fresh background run that continues the work

Compact the current conversation into a handoff note for another session

Save and resume per-project context across Claude Code sessions with simple commands

Check how much context budget is left and how much installed skills consume

Legacy v1 stop hook that extracts session lessons, superseded by continuous learning v2

Learn project-scoped instincts from sessions and evolve them into reusable skills

Report Claude Code token use, spending and budgets by model, session or date

Build MCP servers in TypeScript with tools, resources, prompts and Zod validation

Build, test and deploy Python MCP servers with FastMCP

Complete OAuth by hand for remote MCP servers on headless gateways

List, authenticate and call MCP servers and tools from the terminal

Build high-quality MCP servers in Python or TypeScript

Connect external services with MCP servers configured through plugin files

Build ChatGPT apps with interactive widgets, now superseded by mcp-app-builder

Compare coding agents head to head on your own tasks for pass rate, cost and time

Design agent tool sets, action spaces and observation formats for better completion

Diagnose failed agent runs with capture, diagnosis and recovery reports, not retries

Add x402 payments to agents with per-task budgets and non-custodial wallets

Score each finished task on accuracy, clarity and other axes with evidence

Cut LLM API spend with model routing by task, budget tracking, retries and caching

Build task-local agent harnesses with eval gates and reusable skill extraction

Drive the desktop in the background and escalate only when needed

Drive a real browser to open pages, click through flows and capture screenshots

Launch an AI-controlled Chromium browser with the GStack sidebar extension

Pair a remote AI agent with your browser so it can work in your live session

Import cookies from your Chromium browser into a headless browsing session

Automate browsers over CDP for scraping, testing and screenshots

Reference the Browser Use Cloud API, SDKs, sessions and browser profiles

Show hosted widget documents on a connected macOS panel and control its visibility

Set up a chat channel for OpenClaw using one-line config and secret references

Operate and troubleshoot the OpenClaw Control UI, sessions, dashboards and Gateways

Diagnose OpenClaw Gateway, config, secrets, channel and port failures read-only

Run one-shot Gemini CLI prompts for summaries, generation, hooks and MCP setup

Diagnose failed connections from OpenClaw Control UI and Android, iOS or macOS nodes

Diagnose why an agent session went wrong and draft a bug report for maintainers