The Agent Skills Directory

[COMMAND_EXECUTION]: The skill explicitly forbids itself from writing code, running commands, or creating files. It includes defensive instructions to refuse user requests for direct task execution, ensuring it remains in a strictly advisory role.- [DATA_EXFILTRATION]: The skill reads standard project manifest files (e.g., package.json, go.mod, pyproject.toml) to detect the project technology stack. This information is used solely for local prompt optimization analysis; no unauthorized file access or network activity was found.- [PROMPT_INJECTION]: The skill includes instructions to ignore user attempts to bypass its advisory constraints. While it ingests untrusted data from user prompts and project files (Ingestion points), it is limited to providing recommendations (Capability inventory). No explicit boundary markers or sanitization processes are defined for the ingested data, but the lack of execution capabilities prevents actual harm.

prompt-optimizer