frontend-implement

Pass

Audited by Gen Agent Trust Hub on Jul 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a local Python script python3 workflows/validate_tasks.py to validate the tasks.json configuration file before proceeding with implementation. This is used as a safety check within the project environment.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data from project files such as contracts and task manifests.
  • Ingestion points: Reads .maestro/<slug>/frontend/tasks.json, .maestro/<slug>/lld/frontend.md, and external contract files.
  • Boundary markers: Employs === <path> === delimiters when batch-reading multiple files to maintain context separation.
  • Capability inventory: The skill has the ability to write to the filesystem, edit code, and execute bash commands within the project scope.
  • Sanitization: Includes an automated validation step for the manifest and enforces a human-in-the-loop gate for any task explicitly marked with needs_human_gate: true.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 9, 2026, 09:27 AM
Security Audit — agent-trust-hub — frontend-implement