crawlee
Pass
Audited by Gen Agent Trust Hub on Mar 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands for environment setup, specifically
npm install crawlee playwrightandnpx playwright install chromiumto enable browser-based crawling. - [EXTERNAL_DOWNLOADS]: The skill triggers the download of the
crawleeandplaywrightpackages from the official npm registry, along with browser binaries required for Playwright operations. - [PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by design, as it ingests data from arbitrary external websites.
- Ingestion points: The
requestHandlerimplementation inSKILL.mdextracts text, attributes, and metadata from target URLs. - Boundary markers: The skill documentation lacks specific instructions or markers to differentiate scraped content from agent instructions, increasing the risk of the agent following instructions embedded in website content.
- Capability inventory: The skill utilizes
pushDataandDataset.exportToCSVfor local file writing andenqueueLinksfor further network requests based on scraped data. - Sanitization: There is no evidence of sanitization, filtering, or validation of the ingested external content before it is stored or processed.
Audit Metadata