pseo-strategy
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from external sources, including APIs and web scraping (Phase 2: Data Source Assembly), to generate large-scale content. This creates a surface for indirect prompt injection where malicious data could attempt to influence the agent's output.
- Ingestion points: Phase 2 identifies API data, public datasets, and scraped competitor data as primary inputs.
- Boundary markers: The skill specifies 'Quality gates' and 'Human review' (Phase 5) as checkpoints before deployment.
- Capability inventory: The skill utilizes
scripts/check-output.pyto read and validate files, and instructions indicate the generation of numerous landing pages (file writes). - Sanitization: The process includes 'Duplicate detection' and a '10% sample' human spot-check to verify factual accuracy and quality.
- [COMMAND_EXECUTION]: The skill includes a local utility script,
scripts/check-output.py, used to validate that the generated deliverables contain necessary SEO terms. This script uses standard Python libraries (sys,pathlib) to perform basic text analysis on local files.
Audit Metadata