crawlberg-crawling-a-site
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the crawlberg command-line tool. Examples show the agent executing shell commands with various parameters to crawl domains, manage concurrency, and set rate limits.
- [INDIRECT_PROMPT_INJECTION]: The skill enables the agent to ingest content from arbitrary external websites.
- Ingestion points: External URLs fetched via crawlberg crawl as described in SKILL.md.
- Boundary markers: There are no explicit instructions to wrap or delimit the fetched content to prevent the agent from following instructions found on crawled pages.
- Capability inventory: The skill uses crawlberg for network requests and allows redirection of output to files (e.g., > docs.md).
- Sanitization: The skill converts HTML to Markdown/JSON but does not describe filtering for malicious prompts or hidden instructions within the crawled data.
Audit Metadata