by backnotprop
Agents orchestrating other agents
# Add to your Claude Code skills
git clone https://github.com/backnotprop/orchestratorLast scanned: 7/11/2026
{
"issues": [],
"status": "PASSED",
"scannedAt": "2026-07-11T06:12:38.367Z",
"npmAuditRan": false,
"pipAuditRan": true,
"promptInjectionRan": true
}orchestrator is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by backnotprop. Agents orchestrating other agents. It has 96 GitHub stars.
Yes. orchestrator passed SkillsLLM's automated security scan — a dependency vulnerability audit plus prompt-injection heuristics — with no high-severity issues. You can read the full report in the Security Report section on this page.
Clone the repository with "git clone https://github.com/backnotprop/orchestrator" and add it to your Claude Code skills directory (see the Installation section above).
orchestrator is primarily written in TypeScript. It is open-source under backnotprop on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh orchestrator against similar tools.
No comments yet. Be the first to share your thoughts!
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM is an informational directory and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
Use your harness. To orchestrate many others.
Codex · Claude Code · Copilot CLI · Grok Build · Pi
Orchestrator is a simple skill built on a powerful local CLI inspired by Kubectl. Work with one of your agents; it orchestrates others (Claude Code, Codex, etc) in the background.
You speak in models and outcomes:
Use Fable for the UI, GPT-5.6 Sol for the implementation, and Grok 4.5 for the small fixes. Run independent work in parallel, then have Opus review the result.
Explore the repository with Kimi 2.7. Give the plan to GPT-5.6 Sol, then use Fable only where the interface needs real design work.
npx skills add backnotprop/orchestrator
npm install -g @backnotprop/orchestrator-cli
The CLI install is optional up front. The skill can install it on first use.
/orchestrator launch a gpt5.6 sol agent to build the backend,
a fable agent to design the frontend wireframes,
and after that a grok4.5 agent to build the frontend
Preferences are optional. Edit
skills/orchestrator/PREFERENCES.md, or
tell your agent what to put there:
Use Fable for extensive UI work.
Use GPT-5.6 Sol for deep execution.
Use Grok 4.5 for small coding tasks.
Use Kimi 2.7 for exploration.
When Fable is usage is capped, use GPT-5.6 Sol.
When GPT-5.6 Sol is unavailable, use Opus.
When every preferred model is unavailable, pause and notify me.
Orchestrator discovers current model names and IDs from installed runtimes instead of relying on stale model slugs.
Human model names are fine. The skill resolves them against live runtime catalogs, checks available provider limits when supported, and follows the current request before saved preferences.
Orchestrator is not a harness. You use it in your preferred harness and it orchestrates others in the background.
Every worker is a named task with status, logs, output, follow-up, resume, and stop controls. Agents use a small JSON command loop:
orchestrator doctor --json --compact
orchestrator models <runtime> --json --compact
orchestrator launch <runtime> --json --compact --brief "<task>"
orchestrator ps --json --compact --active --brief
orchestrator read <task-id>... --wait --json --compact
orchestrator interrupt <task-id> --json --compact --reason "<reason>"
The CLI owns process supervision and task state. The calling agent owns judgment, delegation, and synthesis.
Read the Operator Guide for the full control contract, or see custom agents, architecture decisions
Business Source License 1.1 today, with production use allowed except for commercial hosted or managed agent-orchestration services. Converts to Apache License 2.0 on July 9, 2029.