The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard terminal utilities such as find, git ls-files, ls, and awk to perform repository discovery. These commands are used to identify the file structure and validate that configured path patterns match actual files on disk.
[COMMAND_EXECUTION]: The skill facilitates the installation of a local Python script (inject-rules.py) intended to run as a project-specific hook. This script is provided in plain text within the skill, uses only standard Python libraries, and operates locally to ensure rules are correctly loaded when new files are created.
[DATA_EXFILTRATION]: There are no network-related commands or outbound requests. All file reading and writing operations are restricted to the project's .claude/ directory and CLAUDE.md files, with explicit exclusions for sensitive directories like .git and node_modules.
[PROMPT_INJECTION]: The skill's instructions are focused on technical optimization and do not contain attempts to override agent safety guidelines, extract system prompts, or bypass behavioral constraints.
[CREDENTIALS_UNSAFE]: The skill does not attempt to access, store, or transmit sensitive credentials, API keys, or private system configuration files.

optimize-claude-config