pdfops-mcp 0.3.3 → 0.3.5

package/README.md

@@ -61,7 +61,7 @@ Any other stdio client (Windsurf, Cline, Zed, your own agent): command `npx`, ar
 
 A hosted MCP runtime executes this server on a machine where your agent's file paths do not exist. Nothing changes in the config — pass sources the server can reach and skip `output_path`:
 
-- **Inputs** (`pdf_path`, `pdf_paths`): an `https://` URL the server can fetch (≤50 MB), or a `data:application/pdf;base64,…` URI for small files.
+- **Inputs** (`pdf_path`, `pdf_paths`): an `https://` URL the server can fetch (≤4 MB per PDF, ≤4.5 MB per request — the API's limits), or a `data:application/pdf;base64,…` URI for small files.
 - **Outputs**: omit `output_path` and `pdf_fill` / `pdf_merge` / `pdf_invoice` return the PDF inline as an embedded `application/pdf` resource (`pdfops://filled.pdf`, …) that the client saves. With `output_path` set, the file is written where the *server* runs.
 
 Locally, absolute paths keep working exactly as before and remain the recommended form — bytes stay off the model context.
@@ -89,6 +89,6 @@ API docs: [pdfops.dev/docs](https://pdfops.dev/docs) · OpenAPI: [pdfops.dev/ope
 
 ## Privacy Policy
 
-This server runs on your machine and sends only what a tool call needs to the PDFops API (`https://pdfops.dev`): the PDF bytes you point it at, the field values you supply, and your API key if you set one. PDFops processes the request in memory and returns the result; it does not store your documents. Anonymous usage is metered per IP and per client tag (`mcp`) for quota and attribution only. Nothing is shared with third parties. The full policy, including retention and contact details, is at <https://pdfops.dev/privacy>. Questions: hello@pdfops.dev.
+This server runs on your machine and sends only what a tool call needs to the PDFops API (`https://pdfops.dev`): the PDF bytes you point it at, the field values you supply, and your API key if you set one. PDFops processes the request in memory and returns the result; it does not store your documents. API requests are logged with usage metadata to run quotas and the service. The full policy is at <https://pdfops.dev/privacy>. Questions: hello@pdfops.dev.
 
 MIT © PDFops

package/dist/index.js

@@ -82,10 +82,11 @@ server.registerTool('pdf_inspect', {
     outputSchema: {
         count: z.number().int().describe('Number of form fields found; 0 when the PDF has no fillable form'),
         hasXFA: z.boolean().describe('True for hybrid AcroForm/XFA documents, whose XFA layer is dropped when filled'),
+        truncated: z.boolean().describe('True when the form exceeded an API enumeration cap; fields is then a prefix, not the full list'),
         fields: z
             .array(z.object({
             name: z.string(),
-            type: z.string().describe('text, checkbox, radio, optionlist, or unsupported'),
+            type: z.string().describe('text, checkbox, dropdown, radio, optionlist, or unsupported'),
             readOnly: z.boolean(),
             value: z.string().optional(),
             options: z.array(z.string()).optional().describe('Permitted values for radio/dropdown/optionlist fields'),
@@ -175,8 +176,9 @@ server.registerTool('pdf_invoice', {
             invoice_number: z.string().optional(),
             date: z
                 .string()
+                .regex(/^\d{4}-\d{2}-\d{2}$/)
                 .optional()
-                .describe('Shown on the invoice; also pins metadata for determinism'),
+                .describe('Calendar date in YYYY-MM-DD form (other formats are rejected with invalid_date). Shown on the invoice; also pins metadata for determinism'),
             due: z.string().optional(),
             currency: z.string().regex(/^[A-Z]{3}$/).optional(),
             tax_rate: z.number().min(0).max(100).optional(),

package/dist/source.js

@@ -14,8 +14,8 @@
 // Outputs mirror it: with `output_path` the PDF is written to disk; without it
 // the PDF comes back as an embedded application/pdf resource the client saves.
 import { readFile } from 'node:fs/promises';
-/** Upper bound on a fetched or inline PDF. The API rejects larger bodies anyway. */
-export const MAX_INPUT_BYTES = 50 * 1024 * 1024;
+/** Upper bound on a fetched PDF: the API's per-PDF cap, so oversize fails here with a clear message. */
+export const MAX_INPUT_BYTES = 4 * 1024 * 1024;
 export const classifySource = (s) => {
     if (/^data:/i.test(s))
         return 'data';

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pdfops-mcp",
-  "version": "0.3.3",
+  "version": "0.3.5",
   "mcpName": "dev.pdfops/pdfops-mcp",
   "description": "MCP server for the PDFops API — give AI agents deterministic PDF tools: inspect AcroForm fields, fill forms, merge PDFs, and generate invoices. Works with Claude Code, Claude Desktop, Cursor, and any MCP client.",
   "keywords": [