improve-animations
Pass
Audited by Gen Agent Trust Hub on Jul 26, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: A detector hit was triggered by the phrase 'ignore previous instructions' in the SKILL.md file. Detailed analysis confirms this is a defensive 'Hard Rule' (Rule 4) instructing the agent to ignore and report prompt injection attempts encountered within the audited codebase, rather than an attempt to bypass the agent's own safety filters.
- [DATA_EXFILTRATION]: The skill workflow is limited to read-only reconnaissance and analysis of the local repository. There are no network operations, such as 'curl' or 'wget', or instructions to transmit data to external domains.
- [REMOTE_CODE_EXECUTION]: The skill does not download or execute remote scripts. It explicitly prohibits mutating operations, including package installations and builds (Hard Rule 2), ensuring all code analysis remains local and non-executable.
- [COMMAND_EXECUTION]: The skill utilizes standard, safe development tools like 'grep' and 'git' (e.g., 'git rev-parse') for reconnaissance and metadata tracking. These operations are consistent with its stated purpose of auditing a codebase and do not present a security risk.
Audit Metadata