What problem does it solve? Subagent results are injected verbatim into the main conversation context, so verbose agent output quickly exhausts the context window during long sessions. This Skill tells the main thread when to delegate to caveman-compressed subagents instead of doing work inline or using vanilla agents, keeping each delegation's context cost small. ## Core Features & Use Cases - Delegation decision matrix: Maps task types to the right agent — cavecrew-investigator for locating code, cavecrew-builder for surgical 1-2 file edits, cavecrew-reviewer for diff review, or vanilla agents when prose and rationale are wanted. - Structured output contracts: Each subagent returns a compact, greppable format (path:line findings, one-line change summaries, severity-tagged review findings) instead of prose. - Chaining patterns: Supports locate → fix → verify pipelines, parallel investigator scouting, and single-shot edits when the site is already known. - Use Case: During a long debugging session, spawn cavecrew-investigator to find all callers of a function, hand the chosen site to cavecrew-builder for the fix, then have cavecrew-reviewer audit the diff — all while keeping main-context usage low. ## Quick Start Ask the assistant to delegate a code search to the cavecrew investigator subagent so the results come back in compressed form.
