The Agent Skills Directory

[PROMPT_INJECTION] (SAFE): The skill contains no attempts to bypass safety filters. It actually reinforces constraints by explicitly forbidding the agent from coding or implementing changes while the skill is active.- [DATA_EXFILTRATION] (SAFE): There are no network requests, API calls, or patterns suggesting the transmission of sensitive data to external servers.- [REMOTE_CODE_EXECUTION] (SAFE): No scripts, package managers, or shell commands are present. The skill is entirely composed of Markdown instructions.- [INDIRECT_PROMPT_INJECTION] (LOW): The skill possesses a vulnerability surface by directing the agent to review project files and documentation. 1. Ingestion points: Step 1 (Reviewing files/documentation). 2. Boundary markers: Absent. 3. Capability inventory: None (the skill forbids implementation). 4. Sanitization: Absent. The risk is negligible because the skill lacks the tools to act on malicious instructions found in processed data.

brainstorming