ConvertAPI

Convert files, process PDFs, retrieve asynchronous conversion results, and manage temporary files. The integration exposes 17 tools: the 15 existing conversion, PDF, account, discovery and storage tools, plus download_file and delete_async_job.

Connect with an API token or JWT from ConvertAPI authentication settings. The optional Owner or Admin Master Token enables account information. Regular conversion credentials cannot establish account identity through /user; a limited connection label does not verify an account or validate conversion permission. Legacy credentials remain usable when the provider accepts them. A Master Token and conversion token must belong to the intended account; the API does not provide a documented way to prove this binding using an ordinary conversion token.

Choose automatic routing or a documented regional server, including Japan. Regional routing does not establish legal compliance. Tokens are supplied in request headers; JWT expiration and credential rotation remain the account owner's responsibility.

Use list_supported_conversions before converting. It returns current converter inventory with optional parameter metadata. File extensions remain in sourceFormat and destinationFormat; additive converterSourceFormat and converterDestinationFormat identify the actual routes when a single route name exists; plural route arrays preserve multiple native aliases without guessing their pairing. For example, PDF protection produces a PDF while using the protect converter. Discovery is an unpaginated provider inventory; no invented pagination or continuation is applied.

Source files require exactly one public HTTP(S) URL, exact stored file ID, or canonical base64 content with a file name. Existing string parameters and storage defaults are retained. Parameters cannot override source files, storage, authentication, jobs or callbacks. Consult native metadata for format-specific parameters.

Generated files are downloadable whether storeFile is true or false. Stored URLs and file IDs grant access to the content and expire after up to three hours. They cannot be renewed after the provider deletes the data. In-memory results are delivered as files; raw base64 and extracted text bytes are not returned in ordinary result fields. The legacy nullable textContent field remains, always null. conversionTime remains numeric when provided and is omitted when absent; duration and its unit are never invented; current documentation does not establish a unit for this legacy field. The legacy numeric apiKey account identifier is optional because current account responses can omit it.

Specialized PDF tools use the current protect, unprotect, text-watermark, split, compress, merge, txt and pdfa routes. Existing splitByPage, watermark string options, permission booleans and ocrEnabled are translated to the documented parameters. Text watermarking requires watermarkText; image watermarking is not exposed. Explicit OCR options are supported for PDF input; omit them for other formats. PDF/A uses the provider default PDF/A-2b setting and does not prove archival or legal compliance.

Every conversion can consume credits. Deleting a file or job retention record does not refund credits, reverse history or prove processing was cancelled. Deletes distinguish an accepted and verified removal from an already missing or expired resource. Output files and asynchronous job records have separate retention. A request can succeed before its response is lost; reconcile the exact resource before retrying. No file/job inventory or recovery-by-marker endpoint is documented, so ambiguous submissions may require account-owner inspection and automatic retention expiry.

Asynchronous conversion remains available with polling. Processing, completed and missing states follow native HTTP 202, 200 and 404; authentication, conversion and service failures are reported separately. This integration does not expose callbacks or triggers. Undocumented historical converter aliases are not asserted to be retired; generic conversion callers should discover currently supported routes.

See SPEC.md for the capability map and official references. Private controlled verification requires a dedicated account, exact credential hashes, explicit conversion and retained-credit authorization, and independent qpdf/pdftotext observers. No live provider operations were used during this refresh.