The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests external reviewer comments as input ($0). This untrusted data is processed to generate responses. There are no explicit boundary markers or sanitization steps mentioned to prevent instructions embedded within the reviewer comments from influencing the agent's behavior. However, the skill's workflow is limited to text generation, and it lacks dangerous capabilities like network access or shell execution that could be abused via this vector.
[SAFE]: The skill relies on local reference files (rebuttal-prompts.md) for its logic and templates. No network operations, sensitive system file access, or unauthorized command executions were detected.

rebuttal-writing