implementation-review
Pass
Audited by Gen Agent Trust Hub on Aug 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is designed to execute repository-native scripts defined in
package.json. It instructs the agent to: "Use scripts declared in the affected workspace's package.json. Prefer the narrowest meaningful lint, test, typecheck, build, OpenAPI, schema-generation, or integration-test script." and "Derive validation commands from the affected workspace's package.json scripts". This creates a surface for executing arbitrary shell commands if the repository configuration is compromised. - [PROMPT_INJECTION]: The skill processes untrusted input from git diffs, PR descriptions, and commit messages to infer intent and perform reviews. There are no explicit boundary markers or sanitization instructions mentioned to prevent embedded instructions within these external data sources from influencing the agent's behavior (Indirect Prompt Injection).
Audit Metadata