implement-with-subagents

Pass

Audited by Gen Agent Trust Hub on Jul 29, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a local script node scripts/validate-project.mjs during the preflight phase to ensure project integrity. This is a standard automation practice in software development workflows.\n- [PROMPT_INJECTION]: The skill ingests various repository-based data sources, which creates a potential surface for indirect prompt injection.\n
  • Ingestion points: Preflight step 3 involves reading project records, wiki references, affected code, and tests.\n
  • Boundary markers: The skill does not explicitly define delimiters to separate ingested content from its system instructions or provide warnings to ignore embedded commands.\n
  • Capability inventory: The agent has permissions to execute local shell commands (Node.js) and perform repository modifications through Git.\n
  • Sanitization: No explicit sanitization or filtering of the read content is described in the skill instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 29, 2026, 09:29 AM
Security Audit — agent-trust-hub — implement-with-subagents