cavecrew

Delegate code location, editing, and review tasks to subagents with compressed outputs.

Updated May 10, 2026
One-click install
npx skills add https://github.com/AhmadGehad/lfms --skill cavecrew-ahmadgehad
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: cavecrew
Source: https://github.com/AhmadGehad/lfms/tree/main/.kilocode/skills/cavecrew
Command: npx skills add https://github.com/AhmadGehad/lfms --skill cavecrew-ahmadgehad

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This Skill helps manage context exhaustion by delegating specific tasks to subagents, reducing the cognitive load on the main thread and improving efficiency.

Core Features & Use Cases

  • Subagent Delegation: Directly spawn subagents like cavecrew-investigator, cavecrew-builder, or cavecrew-reviewer for specific tasks.
  • Output Compression: Subagent outputs are compressed, reducing the amount of main context required.
  • Use Case: When working on a complex codebase, use cavecrew-investigator to locate code quickly and pass the results to cavecrew-builder for targeted editing, followed by cavecrew-reviewer for a diff audit.

Quick Start

To locate a symbol in the codebase, use the command: 'cavecrew-investigator locate symbol'

Frequently Asked Questions about cavecrew

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

FAQPage Schema
How do I reduce context exhaustion when managing a complex codebase?

Reduce context exhaustion by delegating specific tasks like code location, editing, and review to subagents. This offloads cognitive demand from the main thread, compressing outputs and optimizing overall context efficiency.

How do I use subagent delegation for code review and editing?

Use subagent delegation by spawning subagents like cavecrew-investigator to locate symbols, cavecrew-builder for targeted editing, and cavecrew-reviewer for diff audits. This sequence handles tasks efficiently with compressed outputs.

What is the best way to locate code symbols without loading the entire file context?

The best way to locate symbols without loading full context is using a dedicated investigator subagent. It finds the specific code and passes compressed results to the main thread or builder subagent, preserving context efficiency.

Do I need to manually activate subagents to optimize context efficiency?

Yes, this approach requires subagent activation. You must directly spawn and handle the compressed outputs from these subagents to successfully reduce cognitive load and manage the main thread's context.

What are the limitations of using subagents for task delegation?

A key limitation is the strict requirement for subagent activation and handling compressed outputs. If the environment cannot spawn subagents, this approach fails to reduce cognitive load or manage context efficiently.