run

Orchestrates CLI AI coding agents with unified commands across platforms and execution modes.

15|6|Updated Apr 20, 2026

Search Tags:#automation #workflow #cli #ai #agents #orchestration

Ready to run

npx skills add https://github.com/phnx-labs/agents-cli --skill run-phnx-labs

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: run
Source: https://github.com/phnx-labs/agents-cli/tree/main/skills/run
Command: npx skills add https://github.com/phnx-labs/agents-cli --skill run-phnx-labs

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill solves the fragmentation of CLI coding agents by providing a single, consistent interface to dispatch, manage, and monitor various AI agents across different platforms and environments.

Core Features & Use Cases

Multi-Harness Support: Execute tasks across Claude, Codex, Gemini, Cursor, OpenCode, and more using a unified command syntax.
Flexible Execution Modes: Toggle between interactive TUI sessions, headless automation, and permission-gated execution modes like plan, edit, auto, and skip.
Advanced Observability: Manage complex workflows with features like budget guardrails, automated fallback chains, secret injection, and remote host dispatching.

Quick Start

Use the run skill to execute a headless task with the Claude agent by typing agents run claude followed by your specific prompt.

Frequently Asked Questions about run

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I run multiple AI coding agents from a single command-line interface?▼

You can orchestrate diverse AI coding agents from a single CLI by dispatching tasks through a unified command syntax, supporting multi-harness execution across platforms like Claude, Codex, and Gemini.

What is the best way to automate headless AI agent workflows for coding tasks?▼

Automating headless AI agent workflows is best handled by toggling flexible execution modes that support headless task dispatching, automated fallback chains, and budget guardrails to manage complex development pipelines.

Can I manage AI coding agents on remote hosts and inject secrets during execution?▼

Yes, you can manage AI coding agents on remote hosts by offloading task dispatching to external environments, utilizing built-in secret injection to securely pass credentials during automated workflow execution.

Does this CLI orchestration tool support interactive TUI sessions alongside automated pipelines?▼

Yes, the CLI orchestration tool supports toggling between interactive TUI sessions for hands-on coding and headless automation for pipelines, featuring permission-gated execution modes like plan, edit, auto, and skip.

What are the limitations when dispatching tasks across different AI agent harnesses?▼

While cross-platform agent management is supported, task dispatching depends on the specific permission execution modes and budget guardrails configured, meaning complex workflows require robust observability to prevent runaway token consumption.