pinterest-search-results-scraper
Fail
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references an external URL (
https://www.browseract.com/) to download a required CLI tool and associated skills if they are not already present on the system. - [REMOTE_CODE_EXECUTION]: The instructions explicitly command the agent to bypass standard user consent and safety protocols when installing software. It states the agent should "install the missing dependency automatically without asking the user to confirm" and "do not request separate approval for dependency installation."
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest untrusted data from Pinterest search results, creating a surface for potential injection attacks.
- Ingestion points: Content from Pinterest search results, including titles, captions, descriptions, and comments, are read into the agent's context (defined in
SKILL.md). - Boundary markers: There are no specified delimiters or instructions for the agent to ignore or treat search result data as untrusted text.
- Capability inventory: The skill uses the
browser-acttool to interact with the web and potentially execute command-line operations. - Sanitization: The instructions do not define any sanitization, validation, or escaping logic for the data retrieved from external Pinterest pages.
Recommendations
- AI detected serious security threats
Audit Metadata