# search_qr_codes The search_qr_codes MCP tool finds QR code signatures in a document and returns their decoded text, page, position, and optionally the graphic as a base64 PNG. `search_qr_codes` finds QR codes in a document and returns each one's **decoded text**, page, and position — optionally the graphic itself as a base64 PNG. Filter with `text` to return only codes containing a given string. Example prompt: *"What does the QR code on this invoice say?"* **Tool description (as the AI agent sees it):** > Searches a document for QR code signatures and returns each QR code's decoded text, page, position, and — when returnImage is true — the QR code graphic as a base64-encoded PNG. Supports PDF, DOCX, XLSX, PPTX, and 30+ more document formats. Optionally filters results to QR codes whose decoded text contains a specific string. Do NOT pre-check whether the file exists — pass the filename the user provided directly. Returns a JSON object with `found` (count) and `signatures` (array with `page`, `type`, `text`, position, dimensions, and optional `imageBase64`). On failure, the response text starts with 'QR code search failed for' followed by the underlying exception type, message, and inner-exception chain. ## Parameters | Name | Type | Required | Description | |---|---|---|---| | `file` | object | yes | The document — see [FileInput shape]({{< ref "signature/mcp/tools-reference/_index.md#the-fileinput-shape" >}}) | | `text` | string | no | Return only QR codes whose decoded text contains this string. Omit to return all QR codes. | | `returnImage` | boolean | no | Include the QR code graphic as a base64 PNG in the response. | | `password` | string | no | Password for protected documents. | ## Example call ```json { "name": "search_qr_codes", "arguments": { "file": { "filePath": "invoice.pdf" }, "returnImage": false } } ``` ## Result A JSON array of QR codes: decoded text, page, position, and — when `returnImage` is `true` — the code rendered as a base64 PNG. Leave `returnImage` off unless you actually need the picture: the decoded text is what an agent reasons over, and the image inflates the response considerably. On failure the text starts with `QR code search failed for`, followed by the exception type and message. ## Example prompts * *"What does the QR code on this invoice say?"* * *"Find every QR code containing 'ORDER' and list the pages."* * *"Extract the tracking reference from the QR code."* See it used end-to-end: [Extract QR and barcode data]({{< ref "signature/mcp/use-cases/extract-qr-and-barcode-data.md" >}}).