Agent Launcher Orchestrator
Guides building, launching, grading and scheduling a Claude Managed Agent in your own account.
- Repository
- alirezarezvani/claude-skills
- Stars
- 29,149
- Licence
- MIT
- Files
- 5 files
What it does
This skill helps an AI agent walk you through creating a Claude Managed Agent (an agent that runs on Anthropic's platform under your own account). It keeps a goal file in your project folder, uses small scripts to decide which stage you are in (interview, launch, grading, scheduled running, wrap-up), and can shape the work as a single pass, a capped grade-and-improve loop, or a recurring scheduled job. It produces curl commands you run yourself with your own API key rather than calling the API itself.
Great for
- Developers who want to set up a Claude Managed Agent and need a structured launch process
- Teams that want a rubric-graded agent or a nightly scheduled worker with a clear goal file
- People already comfortable running terminal commands and managing an Anthropic API key
Skip it if
- Non-developers who would have to run curl commands and handle API keys without guidance, since mistakes could incur costs or expose credentials
- Generic agent loops unrelated to Claude Managed Agents; the skill itself points to other skills for those
- Anyone expecting it to write ordinary Claude Code skills (it is scoped to managed agents only)
Try asking
- “Build me an agent that triages my inbox every morning and help me set it up”
- “Grade my agent against a rubric and iterate up to five times until it passes”
- “Set up a nightly worker that runs on a cron schedule in Europe/Berlin time”
- “Launch this as a managed agent and give me the curl command to run”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 5 original files, the MIT licence and an attribution note.
agent-launcher-orchestrator.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip agent-launcher-orchestrator.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip agent-launcher-orchestrator.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Start with a single-sentence goal naming one job; the router refuses vague goals and asks you to split multi-job goals into separate folders
- Keep the loop iteration cap low (3 to 5) and define a concrete rubric before enabling grade-iterate or scheduled runs to control cost
- Review every emitted curl command before running it, and keep your API key out of files and chat logs
What's inside
- README.md
- scripts/goal_router.py
- scripts/goal_state.py
- scripts/loop_compiler.py
- SKILL.md
Redistributed unmodified from alirezarezvani/claude-skills under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.