arkweb-committer-review
Pass
Audited by Gen Agent Trust Hub on Aug 8, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's instructions define a benign persona (ArkWeb Committer) and a structured review process focused on code quality and security. It does not attempt to bypass safety guidelines or override system constraints.
- [DATA_EXFILTRATION]: The skill does not perform unauthorized data transfer. It is designed to read specific project-related files (design documents and source code) and write a review report to a designated project path provided by the user.
- [REMOTE_CODE_EXECUTION]: The skill does not download or execute external scripts, nor does it define any third-party package dependencies.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data (source code and design docs), which constitutes a potential attack surface. However, this is inherent to its primary function as a code reviewer, and the risk is limited as the skill does not have network access or high-privilege execution capabilities.
- Ingestion points: Source code directories and design document paths provided in the task description.
- Boundary markers: None explicitly defined in the prompt to separate data from instructions.
- Capability inventory: Reads files and writes markdown reports to specific repository paths.
- Sanitization: The skill is explicitly instructed to search for security risks (like injections) within the code it reviews, serving as a defensive layer.
Audit Metadata