compose-expert
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill features a 'Review Mode' and an 'Auto-Init' sequence that ingest untrusted data.
- Ingestion points: The skill fetches pull request diffs and full file contents from GitHub using the
ghCLI as documented inreferences/pr-review.md. It also scans local Gradle and Kotlin files inreferences/auto-init.md. - Boundary markers: There are no instructions providing explicit delimiters or 'ignore embedded instructions' warnings for the fetched code content before it is processed by the agent.
- Capability inventory: The skill utilizes shell command execution (
gh,grep,find,xargs,base64) to interact with the environment and remote data (referenced inreferences/pr-review.mdandreferences/auto-init.md). - Sanitization: The instructions do not define any sanitization or filtering protocols for the external code or project metadata being analyzed.
Audit Metadata