The Agent Skills Directory

[SAFE]: The skill defines a set of rigorous thinking and verification steps without any malicious patterns or deceptive intent.
[COMMAND_EXECUTION]: The skill instructs the agent to run skill-gate --strict at the repository root as a verification step. This represents standard development practice for quality assurance within the skill's environment.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied tasks, which is an inherent attack surface. 1. Ingestion points: User requests identifying high-stakes tasks; 2. Boundary markers: Absent; 3. Capability inventory: Local tool execution (skill-gate) and file system access; 4. Sanitization: Absent, although the high-rigor protocol and verification against observed output act as inherent logical safeguards.

overdrive