acp-router

Route plain-language coding requests to Savant ACP runtime sessions or direct acpx sessions.

Updated Aug 23, 2026
One-click install
npx skills add https://github.com/fame0528/Savant-dead --skill acp-router-fame0528
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: acp-router
Source: https://github.com/fame0528/Savant-dead/tree/main/extensions/acpx/skills/acp-router
Command: npx skills add https://github.com/fame0528/Savant-dead --skill acp-router-fame0528

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill streamlines the process of directing plain-language coding requests to the appropriate AI coding environment, whether it's a Savant ACP runtime session or a direct acpx-driven session, preventing manual intervention and ensuring efficient thread management.

Core Features & Use Cases

  • Intelligent Routing: Automatically determines the best environment (Savant ACP runtime or direct acpx) for coding tasks based on user intent.
  • Thread Management: Ensures coding-agent threads are created correctly via sessions_spawn for ACP runtime, avoiding incorrect message actions.
  • Use Case: When you ask Savant to "run this Python script in Claude Code," the acp-router skill will intercept this, identify "Claude Code" as an ACP harness, and initiate the correct session using sessions_spawn with the runtime: "acp" parameter.

Quick Start

Use the acp-router skill to route a request to run code in Gemini CLI.

Frequently Asked Questions about acp-router

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

FAQPage Schema
How do I route plain-language coding requests to different AI coding agents?

To route coding requests, you can use an agent routing tool that intercepts plain-language prompts and automatically directs them to the correct AI coding environment, such as Claude Code or Gemini CLI, using sessions_spawn or acpx execution.

When should I use Savant ACP runtime sessions instead of direct acpx CLI execution?

Use Savant ACP runtime sessions when you need managed coding-agent thread creation via sessions_spawn, and use direct acpx CLI execution when you need direct harness control over environments like Claude Code, Codex, or Gemini CLI.

How does thread management work when spawning ACP runtime coding agents?

Thread management for ACP runtime coding agents works by using sessions_spawn with the runtime set to acp, ensuring new agent threads are created correctly rather than sending incorrect message actions to the harness.

Can I execute code in Gemini CLI by sending a natural language request?

Yes, you can execute code in Gemini CLI by sending a natural language request. An intelligent routing layer intercepts the prompt, identifies the target harness, and initiates the correct acpx-driven session automatically.

What is the best way to manage multiple ACP harness environments for coding assistants?

The best way to manage multiple ACP harness environments is using an intelligent routing mechanism that identifies the target coding assistant from your request and handles the underlying session spawning or direct CLI execution automatically.

Why does my coding agent fail when I try to start a new ACP thread with a message action?

Your coding agent fails because new ACP runtime threads must be created using sessions_spawn with the runtime parameter set to acp, rather than attempting to initiate them through incorrect message actions.