coding
Pass
Audited by Gen Agent Trust Hub on May 20, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or unauthorized data exfiltration behaviors were identified in the skill instructions or templates.\n- [COMMAND_EXECUTION]: The skill manages development operations through subagents. It implements a security best practice by mandating user approval for any NPM package installation, which prevents unauthorized dependency injection.\n- [PROMPT_INJECTION]: The skill features an indirect prompt injection surface as it processes external planning documents. This is mitigated by structural templates and operational boundaries.\n
- Ingestion points: Task definitions and implementation requirements are read from
docs/*-plan.md.\n - Boundary markers: The skill uses
coding-standards.mdas a mandatory prompt template for subagent generation, providing clear constraints.\n - Capability inventory: Subagents are authorized to modify source code files and execute local test suites.\n
- Sanitization: Security relies on the prompt template structure and mandatory compliance checklists included in subagent reporting.\n- [DATA_EXFILTRATION]: No unauthorized data access or network transmission patterns were found. The skill includes explicit instructions to protect local project documentation and prototype directories from deletion.
Audit Metadata