> For the complete documentation index, see [llms.txt](https://docs.docbits.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.docbits.com/advanced-functions-and-tools/docbits-mcp.md).

# DocBits MCP

DocBits exposes a **Model Context Protocol (MCP)** server that lets AI assistants work with your documents, extraction, search, analytics, and configuration — directly from a chat or coding session. Any MCP-compatible client can connect: **Claude Code**, **Claude Desktop**, **Google Gemini CLI**, **OpenAI Codex**, or any custom integration.

The server is mounted on the DocBits API at `/v3/mcp` and is scoped to your organization — every call runs with the permissions of the user (or API key) that connected.

{% hint style="info" %}
**Endpoint — use your region (EU or US):**

* Production (EU) — `https://eu.api.docbits.com/v3/mcp`
* Production (US) — `https://us.api.docbits.com/v3/mcp`

See [Setup & Configuration](/advanced-functions-and-tools/docbits-mcp/setup-and-configuration.md) for the full host matrix and how to sign in with OAuth.
{% endhint %}

## What Can You Do?

With DocBits MCP an assistant can, on your behalf:

* **Find and read documents** — list, search (full-text and fielded), open a document, read its status, logs, page images, and export preview
* **Work on extracted data** — read extracted fields and tables, correct field values, extract tables with AI, validate, approve, and export to your ERP
* **Query the dashboard** — run dashboard queries in natural language, compare search modes, and inspect search capabilities
* **Pull analytics** — executive summaries, success rates, volume trends, top document types, and pending-task detail
* **Handle e-invoices (eDocs)** — list electronic-document types, test and verify e-documents, manage custom extraction attributes
* **Manage master data** — upsert vendor master data, trigger supplier lookups, import supplier-invoice BODs, maintain payment-term and discount mappings
* **Read and change configuration** — organization preferences, document types, field settings, users, and cache
* **Check health** — API and Celery health, current user, and API info

## Tools Overview

DocBits MCP groups well over a hundred tools into the categories below (the exact number grows with every release; your assistant lists the current set). Most mirror existing DocBits REST endpoints, see the [API Reference](/overview-and-basics/readme.md) for request/response detail. An assistant picks the right tool automatically from your request; the tables are a map of what is available.

### Documents & Processing

| Tool                                    | Description                                      |
| --------------------------------------- | ------------------------------------------------ |
| `list_documents`                        | List documents for the organization (filterable) |
| `get_document` / `get_document_by_name` | Open a document by ID or filename                |
| `upload_document`                       | Upload a new document for processing             |
| `get_document_status`                   | Current pipeline status of a document            |
| `get_document_logs`                     | Processing log entries for a document            |
| `query_org_logs`                        | Page through the organization's service logs     |
| `restart_document`                      | Re-run the processing pipeline                   |
| `delete_document`                       | Delete a document                                |
| `get_pending_documents_count`           | How many documents await validation              |

### Extraction & Validation

| Tool                                                | Description                                                                         |
| --------------------------------------------------- | ----------------------------------------------------------------------------------- |
| `get_extracted_fields` / `get_document_fields`      | Read extracted header fields                                                        |
| `get_extracted_tables`                              | Read extracted line-item tables                                                     |
| `get_document_for_validation` / `get_next_document` | Open a document the way the validation screen does; fetch the next one in the queue |
| `get_page_image`                                    | Page image for visual checks                                                        |
| `update_document_fields`                            | Correct or set field values                                                         |
| `validate_document`                                 | Run field validation rules                                                          |
| `approve_document`                                  | Approve a validated document                                                        |
| `validate_and_export`                               | Validate then export to the ERP                                                     |
| `get_export_preview`                                | Preview the export payload before sending                                           |

### Table Extraction

Everything the validation screen and the table-column settings can do with line-item tables. Configuration tools change the organization's document types; extraction and correction tools work on one document; rule tools store what DocBits learned for a supplier.

| Tool                                                          | Description                                                                                                        |
| ------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------ |
| `list_tables` / `get_table_config`                            | Tables and columns of a document type, with the Required / Read Only / Hidden / Use AI flags                       |
| `create_table`                                                | Add a second line-item table to a document type                                                                    |
| `add_table_column` / `delete_table_column`                    | Add or soft-delete a table column                                                                                  |
| `set_ai_table_config`                                         | Switch AI table extraction, vision and structured mode for the organization                                        |
| `extract_table_ai`                                            | Run the AI table extraction on a document (text or vision)                                                         |
| `extract_table_only`                                          | Extract the table without touching the header fields (rules or AI)                                                 |
| `extract_table_nonai`                                         | Rule-based extraction from coordinates you pass, the first step of a table training through the API                |
| `apply_ai_table_formatting` / `apply_table_feedback`          | Reformat an AI table: hide columns, remap, apply user feedback                                                     |
| `remap_table_columns`                                         | Change which extracted column feeds which configured column                                                        |
| `save_table_rules` / `save_table_training_rules`              | Store the coordinate rules or the AI formatting rules for the supplier, so the next document extracts the same way |
| `delete_table_training_rules`                                 | Remove a supplier's saved rules; the next document falls back to the AI                                            |
| `get_table_extraction_report` / `list_table_report_suppliers` | Per-supplier report of how tables are extracted (rules, AI, model), with a preview                                 |
| `compare_table_extraction_models`                             | Extract the same document with two AI tiers and compare the tables                                                 |

### Search & Dashboard

| Tool                                              | Description                                                                                       |
| ------------------------------------------------- | ------------------------------------------------------------------------------------------------- |
| `search_documents` / `search_documents_by_name`   | Full-text and filename search                                                                     |
| `query_document_dashboard`                        | Natural-language dashboard query                                                                  |
| `explain_document_dashboard_query`                | Show how a query was interpreted                                                                  |
| `get_search_capabilities`                         | Which search features the org has enabled                                                         |
| `compare_search_modes`                            | Compare keyword vs. vector results                                                                |
| `diagnose_document_search`                        | Why a document does or does not show up for a given search, classical vs. full-text, side by side |
| `inspect_document_search_index`                   | Whether a document is present in the document, text and vector indices, and when each was written |
| `get_dashboard_columns` / `set_dashboard_columns` | Read or change the columns shown on the document dashboard                                        |

### Touchless Intelligence

Why documents need human involvement and what to change so that fewer do. Every tool here is read-only except `touchless_apply_change`, which writes exactly one previewed setting change.

| Tool                                                             | Description                                                                                                |
| ---------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------- |
| `touchless_get_summary`                                          | Touchless scorecard for the organization: how many documents went through untouched, by period and segment |
| `touchless_list_segments`                                        | Values you can segment the numbers by (supplier, document type, …)                                         |
| `touchless_rank_supplier_opportunities`                          | Suppliers ranked by how much manual work they cause                                                        |
| `touchless_diagnose_supplier`                                    | Why one supplier's documents stop for a human                                                              |
| `touchless_get_issue_cluster`                                    | One recurring issue and the documents it affected                                                          |
| `touchless_get_document_trace`                                   | Full touchless history of one document                                                                     |
| `touchless_explain_pipeline`                                     | Where in the pipeline a document or a cluster stopped                                                      |
| `touchless_explain_po_match` / `touchless_get_po_match_evidence` | Why PO matching succeeded or failed, with the rule set and tolerance maths                                 |
| `touchless_get_recommendations`                                  | What to change, not just what went wrong                                                                   |
| `touchless_list_change_proposals`                                | Setting changes that would raise the touchless rate, with the expected lift                                |
| `touchless_preview_change`                                       | Dry-run one proposal: nothing is written                                                                   |
| `touchless_apply_change`                                         | Apply one previewed proposal to the organization                                                           |
| `touchless_verify_change`                                        | What an applied change turned out to be worth                                                              |

### Analytics & BI

| Tool                       | Description                          |
| -------------------------- | ------------------------------------ |
| `get_executive_summary`    | High-level KPIs for the organization |
| `get_success_rates`        | Touchless / straight-through rates   |
| `get_volume_trends`        | Document volume over time            |
| `get_top_document_types`   | Most frequent document types         |
| `get_pending_tasks_detail` | Detail of outstanding work           |

### Electronic Documents (e-invoices)

| Tool                                                           | Description                                                        |
| -------------------------------------------------------------- | ------------------------------------------------------------------ |
| `list_electronic_document_types`                               | Supported e-invoice formats (ZUGFeRD, XRechnung, Factur-X, UBL, …) |
| `test_electronic_document`                                     | Classify and extract a sample e-document                           |
| `verify_electronic_documents` / `validate_all_edocs`           | Verify e-document extraction                                       |
| `list_edoc_custom_attributes` / `create_edoc_custom_attribute` | Manage custom extraction attributes                                |

### Master Data

| Tool                                                      | Description                                                                     |
| --------------------------------------------------------- | ------------------------------------------------------------------------------- |
| `upsert_vendor_master_data`                               | Create or update vendor master data                                             |
| `trigger_supplier_lookup`                                 | Enrich a document from supplier master data                                     |
| `import_supplier_invoice_bod`                             | Import a supplier-invoice BOD                                                   |
| `get_supplier_invoice`                                    | Read the persisted supplier-invoice (BOD) data by invoice, supplier or document |
| `upsert_payment_term_mapping` / `upsert_discount_mapping` | Maintain calculation mappings                                                   |

### Configuration, Admin & Health

| Tool                                                    | Description                               |
| ------------------------------------------------------- | ----------------------------------------- |
| `get_preferences` / `get_preference` / `set_preference` | Read and change org preferences           |
| `list_document_types` / `get_field_settings`            | Inspect the document-type and field setup |
| `list_users`                                            | List organization users                   |
| `clear_org_cache`                                       | Clear cached configuration                |
| `check_api_health` / `check_celery_health`              | Service health                            |
| `get_current_user` / `get_api_info`                     | Identity and API metadata                 |

## Next steps

* [**Setup & Configuration**](/advanced-functions-and-tools/docbits-mcp/setup-and-configuration.md) — connect Claude Code, Gemini CLI, or Codex (with login or API key)
* [**Examples**](/advanced-functions-and-tools/docbits-mcp/examples.md) — real things customers ask the assistant to do
* [**Electronic Documents (eDocs)**](/advanced-functions-and-tools/docbits-mcp/edocs-mcp.md) — deep dive on the eDoc tools: inspect/test e-invoice processing and manage custom extraction/preview rules (same endpoint, same login)
* [**Workflows & Cards (DocFlow)**](/advanced-functions-and-tools/docbits-mcp/docflow-mcp.md) — manage advanced workflows and partner cards

{% hint style="info" %}
**One MCP for everything.** The eDoc tools and the **Workflows & Cards (DocFlow)** tools run on the **same** `api.docbits.com/v3/mcp` endpoint with the same DocBits login. DocFlow tools appear there with the prefix `docflow_` (for example `docflow_list_workflows`); the DocFlow page documents them under their bare names. Connecting to the DocFlow host directly is still possible but no longer necessary.
{% endhint %}


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.docbits.com/advanced-functions-and-tools/docbits-mcp.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.