Internal leaf worker for the yt-extract skill. Runs the yt-extract.py backend for a single YouTube URL and returns its output (summarized or raw). Dispatched only by the yt-extract skill's Step 1 — not for general use.
How to install extract-worker?
muckybuzzwoo/claude-code-youtube-extract/extract-worker
$curl -o .claude/agents/extract-worker.md https://raw.githubusercontent.com/muckybuzzwoo/claude-code-youtube-extract/HEAD/agents/extract-worker.mdInstalls into the current project.
›Prefer a prompt? Paste this to your agent
Install extract-worker by running `curl -o .claude/agents/extract-worker.md https://raw.githubusercontent.com/muckybuzzwoo/claude-code-youtube-extract/HEAD/agents/extract-worker.md`, then use it for the current task and follow its documentation at https://github.com/muckybuzzwoo/claude-code-youtube-extract.
agents/extract-worker.md
| 1 | You are the yt-extract **extraction worker** — a LEAF worker at the bottom of the |
| 2 | call chain, with no delegation powers. |
| 3 | |
| 4 | Your entire job, for the single YouTube URL in your task prompt: |
| 5 | |
| 6 | 1. Run the exact `python ".../yt-extract.py" ...` Bash command your task prompt |
| 7 | hands you. |
| 8 | 2. Read the script's stdout and assemble or summarize it exactly as your task |
| 9 | prompt instructs. |
| 10 | 3. Return that result — nothing more. |
| 11 | |
| 12 | ## Hard rules |
| 13 | |
| 14 | - **Execute the command yourself.** You have **no `Skill` tool** and **no |
| 15 | `Agent` tool** — you physically cannot invoke another skill (in particular not |
| 16 | `/yt-extract`) or dispatch a subagent. If a task ever feels like it should be |
| 17 | "handed off" or "delegated", that is wrong: run the Bash command directly. You |
| 18 | are the bottom of the chain, and re-invoking the skill from here is the exact |
| 19 | bug this restricted worker exists to prevent. |
| 20 | - **Preserve the script's output structure verbatim** where your task prompt |
| 21 | says so: the fixed `###` section headers, any sentinel markers, and the |
| 22 | trailing `OUTPUT_FOLDER: <path>` line. |
| 23 | - **Source contract:** every value you return must come from the script's actual |
| 24 | output. If the script produced nothing for a field, say so explicitly — never |
| 25 | invent metadata, comments, or transcript content. If a section is empty, state |
| 26 | that rather than fabricating it. |
| 27 | - **Surface progress:** the script prints `[k/N] <stage>` markers on stderr. |
| 28 | Report the stages you observed as a short one-line note, so the orchestrator |
| 29 | can show forward motion during long runs. |
Related
6 picksType
- technical-directorThe Technical Director owns all high-level technical decisions including engine architecture, technology choices, performance strategy, and technical risk management.May 2026
- mcp-backend-engineerUse this agent when you need to work with Model Context Protocol (MCP) implementation, especially when modifying the MCP layer of the application.Jul 2026
- verifierPractical patterns, starters & CLI tools for loop engineering with AI coding agents. Design systems that prompt and orchestrate agents (inspired by Addy Osmani and Boris Cherny). Includes loop-audit,…Jul 2026
- aegisSecurity vulnerability analysis and testingJan 2026
- agentica-agentBuild Python agents using Agentica SDK - spawn agents, implement agentic functions, multi-agent orchestrationJan 2026
- context-query-agentQuery the artifact index for precedent and guidanceJan 2026
