The Agent Skills Directory

[PROMPT_INJECTION]: The skill was evaluated for indirect prompt injection risks. Ingestion points: conversation history and codebase context. Boundary markers: absent. Capability inventory: project issue tracker write access. Sanitization: absent. The design includes a human-in-the-loop confirmation before publication, mitigating the impact of potentially injected instructions.
[DATA_EXFILTRATION]: No exfiltration risk identified. The skill's operations are contained within the repository context and the project's own issue tracker.
[COMMAND_EXECUTION]: No unauthorized or high-risk command execution patterns were found in the skill's instructions or processes.

to-prd