The Agent Skills Directory

[DATA_EXPOSURE_AND_EXFILTRATION]: The skill accesses the hidden directory ~/.claude/projects/ to append telemetry data to skill-telemetry.md. This is a local logging mechanism used for performance tracking and does not involve exfiltrating data to external entities.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external codebase content, creating an attack surface for indirect prompt injection.
Ingestion points: The agent reads various codebase files, data models, and clinical documentation as part of the analysis phases.
Boundary markers: The prompt does not define specific delimiters or instructions to prevent the agent from obeying instructions embedded within the analyzed files.
Capability inventory: The skill's capabilities are limited to local file system read and write operations; it lacks network access or shell execution permissions.
Sanitization: No input validation or content sanitization is performed on the ingested data.

therapy-personalization