The Agent Skills Directory

[DATA_EXFILTRATION]: The skill instructions direct the agent to use Glob and Grep to scan the local codebase for sensitive data signals, including environment variables (.env), API keys, and user schemas. While this is intended to identify service providers like Stripe or Google Analytics for the generated policy, accessing configuration files containing secrets constitutes a data exposure risk.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data from external URLs and local source code to generate legal documentation.
Ingestion points: The skill uses WebFetch to ingest content from user-provided URLs and Read/Grep to ingest local codebase content (SKILL.md).
Boundary markers: There are no explicit instructions or delimiters defined to isolate the scanned content from the agent's core instructions.
Capability inventory: The skill utilizes WebFetch, Read, Write, Grep, and Glob tools.
Sanitization: No sanitization or validation steps are defined for the data retrieved from external websites or local files before it is processed into the final policy.
[COMMAND_EXECUTION]: The skill relies on the execution of filesystem search tools (Grep and Glob) to identify patterns within the user's project, which allows the agent to perform broad reads across the directory structure.

privacy-generator