point-paid-social-template-production

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes a bundled Python script, scripts/preflight.py, to perform mechanical validation of generated PowerPoint decks and PNG images. This script uses standard libraries (zipfile, xml.etree, struct) to inspect file headers and structure for compliance with layout specifications.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data in the form of campaign briefs and visual references.
  • Ingestion points: External campaign briefs and reference images are integrated into the workflow as defined in SKILL.md and references/campaign-fall.md.
  • Boundary markers: The references/source-priority.md file provides clear classification rules (e.g., 'Approved for reuse', 'Reference only', 'Unverified') to help the agent distinguish between authoritative production material and potentially untrusted reference content.
  • Capability inventory: The skill includes capabilities to read and parse local files via scripts/preflight.py and instructions to write/export editable PowerPoint decks and PNG assets.
  • Sanitization: The process relies on manual human-review checkpoints as outlined in references/qa-preflight.md to verify content accuracy and safety before final delivery.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 07:40 PM
Security Audit — agent-trust-hub — point-paid-social-template-production