What problem does it solve? Long coding sessions exhaust the main context window because vanilla subagent results are injected back verbatim as verbose prose. This Skill decides when to delegate to caveman-compressed subagents so each delegation costs roughly one third of the tokens. ## Core Features & Use Cases - Decision Matrix: Maps task types to the right agent — cavecrew-investigator for locating code, cavecrew-builder for edits of 1-2 files, cavecrew-reviewer for diff review, or vanilla agents when prose and rationale are wanted. - Output Contracts: Defines exact structured formats each subagent returns (path:line findings, verified edit summaries, severity-tagged review findings) so the main thread can parse results reliably. - Chaining Patterns: Documents locate-fix-verify and parallel-scout workflows, plus anti-patterns to avoid wasted turns. - Use Case: During a long refactoring session, spawn an investigator to find all call sites of a function, hand the top two sites to the builder for a surgical edit, then have the reviewer audit the diff — all while keeping main-context usage minimal. ## Quick Start Ask the assistant to delegate code exploration to a cavecrew investigator subagent to save context before starting a multi-file investigation.
