improve-animations
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted source code and configuration files from the repository being audited, creating a surface for indirect prompt injection.
- Ingestion points: The agent reads codebase content during the Recon and Audit phases (documented in SKILL.md) to identify motion patterns.
- Boundary markers: Hard Rule 4 in SKILL.md explicitly instructs the agent to treat repository content as inert data and specifically warns to ignore and flag steering instructions such as 'ignore previous instructions'.
- Capability inventory: The agent has permissions to read repository files, write plan files to the 'plans/' directory, spawn subagents for parallel analysis, and execute git commands for metadata collection.
- Sanitization: The skill relies on clear logical instructions and a read-only operating posture rather than automated code sanitization to manage untrusted data.
- [COMMAND_EXECUTION]: The skill utilizes standard utility commands for repository analysis and metadata.
- Evidence: The instructions involve using 'grep' to search for animation-related keywords and 'git rev-parse --short HEAD' to record commit identifiers in the generated plans.
Audit Metadata