quality
Warn
Audited by Gen Agent Trust Hub on Mar 3, 2026
Risk Level: MEDIUMREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill executes 'go test ./...'. In the Go ecosystem, tests are executable code; if the work being reviewed contains malicious tests, they will be executed within the agent's environment during the assessment pipeline.
- [REMOTE_CODE_EXECUTION]: The skill executes 'sh scripts/lint-arch.sh'. As this script is part of the repository under review, it could be modified by a malicious change to execute arbitrary commands when the quality gate is triggered.
- [COMMAND_EXECUTION]: The skill uses system commands including 'git diff', 'git log', 'go vet', and the 'noodle' CLI tool for event emission and session management.
- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it processes untrusted data from the 'cook' phase (source code and git logs) to inform its logic and the messages it emits.
- Ingestion points: Git diffs, commit logs, and the content of changed files.
- Boundary markers: None identified in the prompt instructions.
- Capability inventory: Execution of 'go test', shell script execution, and 'noodle' CLI tool interaction.
- Sanitization: No sanitization or validation of the reviewed content is performed before processing.
Audit Metadata