paw-mkt-pr
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill documentation includes instructions for installing the
agent-browsertool from a trusted organization's repository (github.com/vercel-labs/agent-browser). - [COMMAND_EXECUTION]: The shared patterns and research mode references contain shell commands for environment setup, tool discovery, and running browser-based research sessions. These are intended for the user to configure their local environment.
- [INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests untrusted data from external websites during PR research tasks.
- Ingestion points: Untrusted content is ingested from sources like Google News, Connectively, and LinkedIn using
agent-browserorWebFetch(e.g., inresearch-mode.md). - Boundary markers: The skill does not define explicit delimiters or 'ignore' instructions for the fetched data, focusing instead on extracting specific insights like sentiment and story angles.
- Capability inventory: The skill possesses capabilities to write files (deliverables such as press releases and campaign briefs) and perform network operations via browser automation.
- Sanitization: There is no evidence of specific sanitization or validation of external content before processing.
Audit Metadata