find-animation-opportunities
Pass
Audited by Gen Agent Trust Hub on Aug 24, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill contains an explicit defensive instruction to the agent regarding potential prompt injections found in analyzed source code. The phrase 'ignore previous instructions' was detected, but it is used as a negative pattern for the agent to watch for and disregard when processing external data, rather than an attempt to bypass system safety filters. This is a best-practice defense against indirect prompt injection.
- [DATA_EXPOSURE]: The skill is strictly read-only as defined in its 'Hard Rules'. It is designed to report findings rather than modify files or exfiltrate data. No sensitive file paths or credential patterns were detected.
- [REMOTE_CODE_EXECUTION]: The skill does not perform any remote code execution or external downloads of executable scripts. It references a well-known technical blog (emilkowal.ski) for instructional context, which is a safe, informational reference.
- [COMMAND_EXECUTION]: While the workflow suggests 'sweeping' or using 'grep', these are standard search operations for identifying code patterns and do not involve the execution of untrusted or high-privilege commands.
Audit Metadata