paw-mkt-social
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill audits live URLs and social media profiles (LinkedIn, TikTok, X) to gather marketing insights, creating a surface for indirect prompt injection.
- Ingestion points: Data is ingested from external websites and social media platforms via
agent-browseras described in the research workflows. - Boundary markers: Absent. The instructions do not define delimiters to separate untrusted data from the agent's instructions.
- Capability inventory: The skill can write social media deliverables to the file system and execute shell commands for tool discovery and browser automation.
- Sanitization: Absent. Content retrieved from external sources is not sanitized before processing.
- [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the
agent-browsertool from the Vercel Labs repository to enable automated research capabilities. - [COMMAND_EXECUTION]: The skill uses local scripts (e.g.,
tool-discovery.sh) and CLI commands to manage its execution environment and perform browser-based research.
Audit Metadata