nextjs-reproduce-issue
Pass
Audited by Gen Agent Trust Hub on Jul 27, 2026
Risk Level: SAFE
Full Analysis
- Management of Untrusted Data: The skill processes external content including issue reports, comments, and repository files (ingestion points identified in SKILL.md). It employs clear boundary markers by instructing the agent to treat this data as untrusted and to disregard any instructions that conflict with safety policies.
- Secure Execution Environment: The skill's workflow involves executing reproduction code (capability inventory includes installation and execution via shell). To ensure security, it mandates a disposable, least-privilege environment free from sensitive credentials or host data.
- Sanitization and Verification: Before any execution occurs, the instructions require a thorough inspection of dependency manifests and repository source files (sanitization process), providing a critical checkpoint against malicious payloads.
- Conditional Safety Logic: The agent is instructed to stop and report a blocker if adequate isolation cannot be achieved, prioritizing system integrity over task completion.
Audit Metadata