The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's core functionality relies on executing subcommands of the caveman CLI (e.g., learn report, learn apply, mem remember) to analyze token usage and manage memory-offloading tasks.
[DATA_EXPOSURE]: To perform its optimization tasks, the skill reads from local project files such as CLAUDE.md and AGENTS.md, and accesses temporary candidate files in ~/.caveman/candidates/. It employs SHA256 hashing to verify that the file content matches the analyzed state before applying any modifications.

caveman-learn