cavecrew

Delegate locate, edit, and diff review tasks to compressed caveman-style subagents.

1|Updated Apr 27, 2026
One-click install
npx skills add https://github.com/dwaynimay/IoTProject --skill cavecrew-dwaynimay
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: cavecrew
Source: https://github.com/dwaynimay/IoTProject/tree/main/.agents/skills/cavecrew
Command: npx skills add https://github.com/dwaynimay/IoTProject --skill cavecrew-dwaynimay

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It helps you decide when to delegate coding and review work to specialized caveman-style subagents, avoiding context bloat while still getting useful, structured results back.

Core Features & Use Cases

  • Delegation decision guide: Chooses the right subagent based on task scope (locate, build/edit, or review).
  • Structured output contracts: Ensures returned results are file-path-first and verifiable, suitable for quick follow-up.
  • Token-saving behavior: Uses caveman-compressed tool results so main context lasts longer across long sessions.

Quick Start

Say: Use cavecrew to delegate this work and compress the subagent outputs back into main context.

Frequently Asked Questions about cavecrew

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

FAQPage Schema
How do I reduce context bloat when delegating code navigation and editing tasks to subagents?

You can reduce context bloat during code navigation by delegating work to caveman-style subagents that return compressed, file-path-first outputs, keeping your main context window clear for longer coding sessions.

What is the best way to manage subagent workflows for surgical code changes within a small file scope?

Managing subagent workflows for surgical code changes is best handled by selecting a builder preset, which applies structured output contracts to ensure returned edits are verifiable and concise without generating long prose.

How do I delegate diff review tasks without filling up my main context window?

To delegate diff review without filling up your context window, use a specialized reviewer subagent preset that analyzes changes and returns compressed, structured results instead of verbose text explanations.

Can I use compressed subagent outputs for locating issues across a large codebase?

Yes, you can use an investigator subagent preset for codebase navigation to locate issues, returning structured, file-path-first results that make follow-up actions quick while saving tokens in the main session.

Does delegating tasks to structured subagents work without extensive setup or external dependencies?

Delegating tasks to structured subagents works without external dependencies, requiring only the selection of the correct preset—investigator, builder, or reviewer—based on your specific task scope to enforce output contracts.