compose-performance-audit
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted Kotlin source code and Android profiling traces (Layout Inspector, Perfetto) to provide optimization advice. This creates a potential surface for indirect prompt injection via malicious code comments or structures designed to manipulate the agent's diagnostic output.
- Ingestion points: Section 1 and 2 of SKILL.md involve collecting target Composable code and profiling exports.
- Boundary markers: The skill does not define specific delimiters or instructions to isolate untrusted data from the analysis logic.
- Capability inventory: The skill is strictly diagnostic and informational; it does not request permissions for file system writes, network communication, or shell command execution.
- Sanitization: No sanitization of user-provided code or data is specified.
- [SAFE]: The skill follows established Android development best practices and references official documentation from Google. No indicators of credential theft, remote code execution, or persistence were found.
Audit Metadata