Skill UI

Cavecrew provides a set of specialized subagents (investigator, builder, reviewer) designed for efficient code development delegation. Unlike vanilla agents, Cavecrew compresses output results, drastically reducing the token usage injected back into the main context. This allows users to maintain context over much longer, complex coding sessions, effectively preventing context exhaustion during large-scale refactoring, debugging, or multi-step feature development.

This utility displays the precise, real-time token usage and estimated cost savings for the current session. It reads directly from the internal session log, ensuring objective and accurate data without requiring AI computation. Triggered by /caveman-stats, it provides immediate performance metrics to monitor API costs and usage patterns.

This skill analyzes and reports on the token usage, expenditure, and budget for AI models (like Claude Code) by querying a local SQLite database. It is essential for users who need detailed financial accountability, tracking costs by project, tool, or date, and managing operational budgets.

This skill automatically routes subtasks to the most cost-effective Claude model (Haiku, Sonnet, or Opus) based on complexity. It logs real token usage and cost per session, and includes an A/B testing feature to validate model tier savings against your actual workload, ensuring accurate cost control.

This advanced skill manages complex, large-scale development tasks by coordinating multiple, specialized subagents in parallel. It systematically decomposes the goal into isolated missions, manages resource quotas (token usage), and handles error recovery across multiple files and components. Ideal for large feature builds, system refactoring, or complex applications requiring multiple, concurrent roles (Planner, Builder, Tester).