paw-mkt-psychology

Pass

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches the agent-browser tool from the Vercel Labs GitHub repository for automated research tasks.
  • [COMMAND_EXECUTION]: The pre-flight protocol involves executing local shell scripts such as tool-discovery.sh and chrome-profiles.sh within the vendor's directory to verify environment configuration.
  • [EXTERNAL_DOWNLOADS]: Instructions are provided to install the agent-browser package and Playwright Chromium via npm and npx during the setup phase.
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it is designed to audit live website URLs and fetch external content for analysis. 1. Ingestion points: Browser automation fetching content from live URLs (referenced in references/shared-patterns.md and SKILL.md). 2. Boundary markers: Absent; there are no specific markers used to distinguish between untrusted web content and system instructions. 3. Capability inventory: Browser interaction via agent-browser, local file system writes for campaign deliverables, and shell command execution for setup. 4. Sanitization: No explicit sanitization or filtering logic is defined for processing external site content.
Audit Metadata
Risk Level
SAFE
Analyzed
May 20, 2026, 10:38 AM
Security Audit — agent-trust-hub — paw-mkt-psychology