The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a local script located at $CLAUDE_PLUGIN_ROOT/scripts/file_metrics.py to determine file size and type. This is a functional requirement for the skill's routing logic.
[EXTERNAL_DOWNLOADS]: The skill references internal template files and documentation (e.g., ../summarizer/templates/ and ../summarizer/references/). These are local filesystem references intended for configuration and do not involve remote network requests.
[DATA_EXFILTRATION]: The instructions specifically cover the processing of configuration files, including sensitive targets like .env and credentials. The skill includes an explicit safety instruction: "note presence, do not expose values," which mitigates the risk of credential leakage in the generated summaries.
[SAFE]: The skill implements a "quote-grounding" technique to ensure summaries are faithful to the source text, which helps reduce hallucinations and maintains high fidelity when summarizing user-provided content.

file-summarization