intent-data
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from multiple sources including user-pasted exports, LinkedIn profiles, and review sites (G2, TrustRadius). This represents a significant attack surface where an attacker could embed malicious instructions in external content. 1. Ingestion points: SKILL.md mentions reading LinkedIn profiles, company sites, and review sites via Playwright. 2. Boundary markers: references/agent-security.md provides explicit instructions to treat untrusted content as data and never instructions, quoting any detected injection attempts. 3. Capability inventory: The skill uses Playwright for web browsing and Intempt MCP for data access. It also writes to .agents/product-context.md. 4. Sanitization: The skill instructs the agent to report injection attempts as strings and not follow URLs found within untrusted content.
- [COMMAND_EXECUTION]: The skill utilizes the Playwright tool to browse external websites. While intended for research, this capability could be abused if the agent is successfully injected. However, specific instructions are included to prevent login/password submission and to treat page content as data only.
Audit Metadata