The Agent Skills Directory

[SAFE]: The skill instructions do not contain malicious patterns such as prompt injection, obfuscation, or data exfiltration. It directs the agent to interact with the user and codebase in a standard, secure manner.\n- [SAFE]: Evaluation of potential indirect prompt injection surface: 1. Ingestion points: User-provided plans, design descriptions, and local codebase files referenced during the session (SKILL.md). 2. Boundary markers: Absent. 3. Capability inventory: Local codebase exploration tools. 4. Sanitization: Absent. The risk is assessed as safe as the functionality is limited to an interactive session concerning the user's own data.

grill-me