cavecrew

Route programming tasks to cavecrew subagents for codebase navigation, edits, and reviews.

Updated Mar 11, 2026
One-click install
npx skills add https://github.com/selfagency/agentsy --skill cavecrew-selfagency
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: cavecrew
Source: https://github.com/selfagency/agentsy/tree/main/.agents/skills/cavecrew
Command: npx skills add https://github.com/selfagency/agentsy --skill cavecrew-selfagency

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It helps you decide when to delegate code-finding, small edits, or diff reviews to specialized subagents without bloating your main context with long prose.

Core Features & Use Cases

  • Delegation decision guide: Choose between cavecrew-investigator, cavecrew-builder, and cavecrew-reviewer based on task type and scope.
  • Context-saving compressed outputs: Ensures subagent tool-results are injected back into the main context in a more compact, caveman-compressed format.
  • Chaining patterns: Supports common workflows like Locate → fix → verify, plus parallel scouting for broader investigations.

Quick Start

Tell the AI: "Use cavecrew to delegate this task: find where X is defined, then make a ≤2 file surgical edit, and finally review the resulting diff."

Frequently Asked Questions about cavecrew

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

FAQPage Schema
How do I delegate code navigation and editing tasks to subagents without bloating context?

Agent delegation routes programming tasks to specialized subagents, ensuring code navigation and small scoped edits are handled outside the main context, with results injected back in a compact caveman-compressed format.

What is the best way to structure subagent outputs for diff review during agentic development?

Structured diff review enforces output contracts for paths, line ranges, severity, and totals. This deterministic routing keeps subagent tool-results compact and prevents verbose prose from consuming main-context usage.

When should I use subagent delegation instead of inline code editing?

Subagent delegation applies to codebase navigation, small scoped edits under two files, and structured diff reviews. It determines when to route work to investigator, builder, or reviewer roles instead of doing work inline.

Can I chain multiple subagent roles for a locate, fix, and verify workflow?

Chaining patterns support common workflows like Locate → fix → verify, plus parallel scouting for broader investigations. You can delegate code-finding, surgical edits, and diff reviews across specialized subagent roles sequentially.

Does context compression for subagents require any external dependencies?

Context compression requires no external dependencies. It applies deterministic routing to cavecrew subagent roles and enforces structured output contracts to keep main-context usage compact during agentic development sessions.