paw-mkt-pr

Pass

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill utilizes browser-based research to ingest data from external sources such as Google News and social media platforms. This activity presents a surface for indirect prompt injection where malicious instructions embedded in third-party web content could potentially influence the agent's behavior during the research process.
  • Ingestion points: Web research commands defined in references/research-mode.md.
  • Boundary markers: Absent from the provided instructions and research patterns.
  • Capability inventory: Browser automation via agent-browser and the ability to save deliverables to the local .pawbytes/ directory.
  • Sanitization: Not explicitly implemented for the ingested web content.
  • [EXTERNAL_DOWNLOADS]: The skill provides instructions to download and install the agent-browser tool directly from Vercel Labs' official GitHub repository for research purposes.
Audit Metadata
Risk Level
SAFE
Analyzed
May 20, 2026, 10:37 AM
Security Audit — agent-trust-hub — paw-mkt-pr