The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill is explicitly instructed to scan sensitive configuration files such as .env, .env.example, .env.sample, and .env.template specifically to find and identify secret keys like STRIPE_SECRET_KEY, SENTRY_DSN, RESEND_API_KEY, and TRIGGER_SECRET_KEY.
[DATA_EXFILTRATION]: The workflow requires reading a broad range of sensitive application files, including dependency manifests (package-lock.json, yarn.lock), infrastructure configs (docker-compose.yml), and source code across various languages, which may contain proprietary information or credentials.
[PROMPT_INJECTION]: The skill exhibits a significant surface for Indirect Prompt Injection by ingesting and processing untrusted data from the project being scanned.
Ingestion points: Project files including package.json, requirements.txt, .env, Dockerfile, and various source code files.
Boundary markers: None specified; the agent is not instructed to ignore or delimit embedded instructions within the scanned data.
Capability inventory: The skill performs extensive file reading across the project directory and has the capability to write a new file (DPA.md).
Sanitization: No evidence of sanitization, escaping, or validation of the content read from external project files before it is processed for DPA generation.

gdpr-dpa