subagent-cursor

Route local Cursor subagent calls and models for bounded Codex companion workflows.

Updated Apr 23, 2026
One-click install
npx skills add https://github.com/Ocean326/Agents --skill subagent-cursor
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: subagent-cursor
Source: https://github.com/Ocean326/Agents/tree/main/skills/global/subagent-cursor
Command: npx skills add https://github.com/Ocean326/Agents --skill subagent-cursor

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

Provide a bounded, local external subagent workflow by routing calls to a local Cursor instance via cursor-agent, enabling safe comparison, quick passes, and delegated tasks without embedding Cursor directly into the main agent.

Core Features & Use Cases

  • External subagent orchestration: route and execute Cursor-based prompts as a bounded companion to Codex workflows.
  • Model routing and evaluation: switch between Cursor models and safe defaults for controlled experimentation.
  • Quick handoff and isolation: attach skill guidance and run independent passes without affecting the primary agent.

Quick Start

Invoke the run_cursor_subagent script with a workspace, model, and prompt to begin a bounded external subagent pass.

Frequently Asked Questions about subagent-cursor

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

FAQPage Schema
How do I route local Cursor as an external subagent for Codex workflows?

Routing local Cursor as an external subagent involves executing the run_cursor_subagent script with a defined workspace, chosen model, and prompt to process bounded sidecar tasks independently alongside your primary Codex workflows.

What is a bounded external subagent and when do I need one?

A bounded external subagent is an isolated local Cursor instance routed via cursor-agent to execute independent prompts, enabling safe model comparison and delegated tasks without affecting your primary Codex workflows.

Do I need cursor-agent available to use local model routing for sidecar tasks?

Yes, cursor-agent availability is required. You must also define a workspace, select a model, and specify a mode (plan or ask) to properly execute local model routing for bounded sidecar tasks.

Can I attach skill files to inline guidance for independent evaluation passes?

Yes, you can attach optional skill files when invoking the run_cursor_subagent script to inline guidance and execute independent evaluation passes without impacting the primary agent.

What is the best way to switch between Cursor models for controlled experimentation?

The best way to switch Cursor models for controlled experimentation is routing calls through a bounded external subagent, applying safe defaults and specified plan or ask modes to isolate evaluation passes.