CoreClaw Skill
The CoreClaw Skill packages the full CoreClaw OpenAPI v2 workflow as an AI-agent skill — discover Workers, inspect input schemas, run Workers and tasks, poll runs, fetch results, export data, and read logs, all from inside your AI assistant. It is MCP-first and v2-only.
It ships as a Claude Code plugin (the recommended path) and can also be used as a standalone skill or a Codex skill.
What you can do
Section titled “What you can do”- Discover Workers in the Store or your private workspace.
- Inspect a Worker’s input schema before running it.
- Run a Worker or a saved task (async by default), with batch runs supported.
- Poll & verify —
poll_runwaits for completion;verify_runreturns a structured verdict. - Retrieve results — paginated rows, or export as CSV/JSON (eight formats, default
csv). - Read logs with in-process
grepfiltering, plus reruns and aborts.
The skill exposes 37 tools — the 34 OpenAPI v2 operations plus three orchestration helpers (poll_run, verify_run, run_workers_batch). It calls the hosted MCP endpoint (https://mcp.coreclaw.com/mcp) and falls back to the REST API (https://openapi.coreclaw.com) with your CORECLAW_API_KEY.
Prerequisites
Section titled “Prerequisites”- Claude Code installed
- A CoreClaw account with an API key from Console → Settings → API & Integrations
Install as a Claude Code plugin (recommended)
Section titled “Install as a Claude Code plugin (recommended)”Two commands:
claude plugin marketplace add Core-Claw/CoreClaw-Skillclaude plugin install coreclaw@coreclaw-skillThen set your API key as an environment variable:
export CORECLAW_API_KEY=YOUR_CORECLAW_API_KEYOnce installed, the skill is available under the /coreclaw-skill:coreclaw namespace in Claude Code.
Use it
Section titled “Use it”Ask Claude Code to do CoreClaw work in natural language:
Find a Google Maps worker on CoreClaw, inspect its input schema, run it for coffee shops near Times Square, and return the name, address, and rating for the top 10.
Claude Code will call list_store_workers → get_worker_input_schema → run_worker → poll_run → list_worker_run_results on your behalf.
Other installation modes
Section titled “Other installation modes”- Standalone skill — clone the repo into
~/.claude/skills/coreclaw; it loads as/coreclaw. - Codex Desktop — symlink
skills/coreclawfrom the repo into Codex’s skills directory.
See the CoreClaw-Skill README for full details.
Next steps
Section titled “Next steps”- MCP Server overview — connect other AI clients (Cursor, Codex, ChatGPT, …)
- CoreClaw API documentation