nerd-execute
Pass
Audited by Gen Agent Trust Hub on Aug 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill implements a disciplined workflow that requires a resolved 'Focus Record' before any mutations occur, which scopes agent behavior. It also includes explicit instructions to never store raw code, shell strings, secrets, permissions, or action authority in its memory systems, representing a proactive data security measure.
- [COMMAND_EXECUTION]: The skill instructs the agent to run a local script (python3 scripts/validate_skills.py) following any modifications to the skill itself. This serves as a self-testing mechanism to ensure the integrity of the skill's logic within the repository context.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from the user's repository, including adjacent code and configurations, creating a surface for indirect prompt injection. Ingestion points: Repository context inspection (Adjacent callers, callees, configuration, and dependencies). Boundary markers: The requirement for a resolved 'Focus Record' from a separate planning step acts as a constraint. Capability inventory: The skill has the authority to mutate repository files and execute a local validation script. Sanitization: The skill relies on structured execution mappings and pre-planning gates rather than explicit content sanitization.
Audit Metadata