build-daily-inspiration-sites

Pass

Audited by Gen Agent Trust Hub on Aug 25, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses a local Node.js script (scripts/validate_capture.mjs) to validate manifest and content files. The script performs file existence and size checks on paths resolved relative to the user-provided manifest location.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core functionality of ingesting and processing external inspiration captures.
  • Ingestion points: The agent reads and processes manifest.json and content.md files from dynamically identified article directories within the articles/ workspace.
  • Boundary markers: The instructions explicitly warn the agent to use source content as "evidence and art direction" only, not as a template, and to never reuse specific brand assets or copy.
  • Capability inventory: The skill allows for local script execution, file system access for validation, and the creation of new sub-tasks/threads containing data derived from the external sources.
  • Sanitization: The instructions require the agent to generate an "originality matrix" to ensure all source identity, pricing, claims, and imagery are replaced with original content before sub-tasks are initialized.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 25, 2026, 03:09 AM
Security Audit — agent-trust-hub — build-daily-inspiration-sites