DocuPipe MCP server
OfficialServer Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| DOCUPIPE_API_KEY | Yes | Required. Your DocuPipe API key. | |
| DOCUPIPE_BASE_URL | No | Optional. Overrides the API address. Defaults to https://app.docupipe.ai. | https://app.docupipe.ai |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
DocuPipe turns documents (PDF, scans, images, Word, Excel, email) into structured JSON. For 'pull X out of these files', call extract once per file with a json_schema of the fields you need, then reuse the returned schema_id for the remaining files so every result has the same fields. Long jobs return status 'processing' with a 'next' call to resume; never re-upload to retry.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| extractA | One call from a file to structured JSON: uploads a local file or URL (or reuses a document_id), runs a DocuPipe extraction, waits and returns the extracted data. Pass schema_id to reuse a saved schema, json_schema to define the fields inline, or neither to let DocuPipe draft a schema from the document (its schemaId comes back so later calls give the same fields). If it returns status 'processing', call the tool in 'next' to resume; it never uploads or extracts twice. Credits: upload 1/page for PDF, image, Word and HTML (CSV, Excel, JSON, text and XML upload free); extraction 2/page at standard effort, 4/page at high effort, 3/page with citations (always standard effort). Without a schema, drafting one costs 1 credit per page on at most the first 5 pages. Creating a schema from json_schema is free. |
| upload_documentA | Upload a local file or URL to DocuPipe and wait for it to be read (OCR and parsing). Returns the document_id used by every other tool. Credits: 1/page for PDF, image, Word and HTML; CSV, Excel, JSON, text and XML upload free. |
| get_documentA | Fetch a document's processing status, page count and parsed text (tables as markdown). Set include_layout for per-page sections with bounding boxes. Free. |
| standardizeA | Run an extraction (standardization) on one uploaded, fully processed document and wait for the result. Use extract instead when starting from a file. Credits: upload 1/page for PDF, image, Word and HTML (CSV, Excel, JSON, text and XML upload free); extraction 2/page at standard effort, 4/page at high effort, 3/page with citations (always standard effort). Without a schema, drafting one costs 1 credit per page on at most the first 5 pages. Creating a schema from json_schema is free. |
| get_standardizationA | Fetch the extracted JSON of a finished standardization. While the extraction is still running this answers not found; wait on its job first. Free. |
| create_schemaB | Save a JSON Schema as a DocuPipe schema so every extraction returns the same fields. Free. Use type object with a description on each property; arrays of objects for line items. |
| list_schemasA | List the account's saved schemas (id and name), newest first, to reuse one instead of creating a new one. Free. |
| list_classesA | List the account's classification classes (id, name, description). classify needs class ids from here. Free. |
| classifyB | Classify one uploaded document into the account's classes and wait for the answer. Credits: 0.1/page, rounded up per document (at least 1). |
| split_documentA | Split one uploaded document into sub-documents, with AI deciding where each one starts, and wait for the new document ids. Credits: 0.2/page, rounded up. |
| list_workflowsA | List workflows that run_workflow can start on one document: classic workflows and Architect workflows that are runnable and take a single document. Free. |
| run_workflowA | Upload a local file or URL with a workflow attached and, for Architect workflows, wait for the run's outputs. Classic workflows return the ids of the jobs they started (use wait_for_jobs, then get_standardization). Credits: the upload plus each step the workflow runs. Resume an Architect run with workflow_run_job_id. |
| convert_to_excelA | Turn a document into an .xlsx workbook without a schema (one sheet per table). Takes a local file, URL or document_id, waits for the workbook and can save it locally. Credits: upload, plus 2/page for the conversion. Resume with the call in 'next'. |
| get_excel_conversionA | Check an Excel conversion, wait for it to finish, and optionally save the .xlsx locally. The download link is fresh on every call. Free. |
| get_jobA | Fetch one job's status, type, error message and results (for split, classify and workflow jobs). A job in error is returned as a tool error. Free. |
| wait_for_jobsA | Wait until the listed jobs finish (any job type) and report each one's status. Returns early when all are done or one fails; call again if some are still processing. An id that stays 'unknown' may be wrong or belong to another account (check it with get_job). Free. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 16 tools
Most tools target distinct steps of the pipeline, and descriptions explicitly resolve the main overlaps (extract vs upload_document+standardize, get_standardization vs get_job/wait_for_jobs). Some residual ambiguity remains among the result-fetching tools (get_standardization, get_job, wait_for_jobs, get_excel_conversion), which all 'retrieve job output' in slightly different ways. An agent can still pick correctly by following the stated sequencing guidance.
All names are lower snake_case, which is consistent and readable. The pattern is mostly verb_noun (upload_document, get_document, list_schemas, run_workflow), with a few bare verbs (extract, standardize, classify) that lack the resource noun but remain unambiguous. Minor deviation only.
16 tools for a document-extraction platform covering upload, extraction, schemas, classification, splitting, workflows, Excel conversion and job polling is reasonable and each tool maps to a distinct capability. It sits just above the ideal 3-15 range, and a couple of job-polling tools could arguably be folded together. No obvious filler tools.
The end-to-end extraction lifecycle is well covered: upload, extract, retrieve text/results, schemas, classify, split, workflows, Excel export and job waiting. Gaps are minor: no delete_document or delete/update schema, and classes are read-only (list_classes with no create_class), but these are manageable via the web console rather than blocking agent workflows.