AI agents call crawl to retrieve information from Crawl4ai without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves and extracts web content through crawling, which is a read-only operation. It has no side effects on the target systems or data—it merely traverses and collects information. The severity is low because unauthorized web scraping may violate terms of service or rate limits, but it does not delete, modify, execute arbitrary code, or move funds.
From the tool's definition Tool performs 'breadth-first crawl' and 'returns markdown per page' from a seed URL—a retrieval and querying operation with no modification or deletion of data. The description explicitly states the output is informational (markdown).
Attacks that exploit this kind of access
Breadth-first crawl up to max_depth starting from seed_url. Returns markdown per page by default. It is categorised as a Read tool in the Crawl4ai MCP Server, which means it retrieves data without modifying state.
Register the Crawl4ai MCP server in PolicyLayer and add a rule for crawl: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Crawl4ai. Nothing to install.
crawl is a Read tool with low risk. Read-only tools are generally safe to allow by default.
Yes. Add a rate_limit block to the crawl rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for crawl. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
crawl is provided by the Crawl4ai MCP server (sadiuysal/crawl4ai-mcp-server). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Every MCP server has a record like this.
Type a name, get the same breakdown: verified identity, auth posture, risk grade, capabilities, recommended policy.
Teams ship this data inside their own products. See what a licence covers →