AI agents call ocr_pdf to retrieve information from Ocr without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
OCR extraction is fundamentally a read operation: it retrieves/extracts text content from PDF documents without modifying, deleting, or executing anything. Even though the tool description is empty, the context (server purpose and sibling tool names) makes the function clear.
From the tool's definition Tool name 'ocr_pdf' and sibling tools 'ocr_image', 'batch_ocr', 'evaluate_accuracy', 'list_engines', 'compare_engines' all indicate data retrieval and analysis operations.
Attacks that exploit this kind of access
ocr_pdf. It is categorised as a Read tool in the Ocr MCP Server, which means it retrieves data without modifying state.
Register the Ocr MCP server in PolicyLayer and add a rule for ocr_pdf: 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 Ocr. Nothing to install.
ocr_pdf 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 ocr_pdf 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 ocr_pdf. 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.
ocr_pdf is provided by the Ocr MCP server (prekzursil/abbyy-finereader-ocr-mcp). 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 →