hermes-acp-orchestrator

Route delegated engineering tasks to Hermes, Codex, and Claude Code with isolated contexts.

Updated Aug 23, 2026
One-click install
npx skills add https://github.com/Rainhoole/hermes-agent-acp-skill --skill hermes-acp-orchestrator
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: hermes-acp-orchestrator
Source: https://github.com/Rainhoole/hermes-agent-acp-skill/tree/main
Command: npx skills add https://github.com/Rainhoole/hermes-agent-acp-skill --skill hermes-acp-orchestrator

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It standardizes and safeguards ACP-style multi-agent delegation so delegated work is routed to the right agent without contaminating the parent context.

Core Features & Use Cases

  • Explicit agent routing to hermes, codex, or claude-code for predictable delegation.
  • Context isolation so delegated tasks don’t pollute the parent turn.
  • Safety guardrails including timeout- and output-size limits for external agents like Codex and Claude Code, suitable for engineering workflows.
  • Use case example: delegate a risk review to claude-code, a patch and tests to codex, and a final integration summary back to hermes.

Quick Start

Ask the assistant to delegate a batch of tasks with explicit agent targets for risk review, patch-and-test implementation, and final integration summarization while enforcing external agent timeouts and output limits.

Frequently Asked Questions about hermes-acp-orchestrator

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

FAQPage Schema
How do I delegate engineering tasks to multiple AI agents without polluting my main context?

Multi-agent delegation isolates delegated context by routing tasks explicitly to specific agents like Codex or Claude Code, preventing delegated work from contaminating the parent turn.

Can I route specific coding tasks to Codex and risk reviews to Claude Code in the same workflow?

Yes, explicit agent routing allows you to assign a risk review to Claude Code, a patch and tests to Codex, and a final integration summary to Hermes within a single batch execution.

How do I enforce execution timeouts and output limits when delegating to external coding agents?

Timeout-aware execution and external output-size guardrails enforce bounded results for external agents, ensuring delegated tasks return concise parent-turn summaries without exceeding limits.

What is context isolation in multi-agent orchestration and when do I need it?

Context isolation in multi-agent orchestration prevents delegated subagent outputs from polluting the parent context, needed for engineering workflows requiring mixed-agent batch execution like implementation and synthesis.

Does multi-agent delegation work for batch execution of implementation, review, and synthesis tasks?

Multi-agent delegation supports engineering workflows requiring mixed-agent batch execution, explicitly routing implementation, review, risk analysis, and synthesis tasks to appropriate agents with bounded results.