PaperOffice Document Operations

The AI document layer for your agent — 343 tools for business documents, EU-hosted.

От сообществаРаботаетБез входаГлобальныйБесплатноТолько чтение

Что умеет

  • Po Documents Search: READ-ONLY. Returns data, changes nothing. po_documents_search: Stable document search for keyword, filename and available metadata. Guaranteed: validates required query/scope inpu
  • Po Documents Get: READ-ONLY. Returns data, changes nothing. po_documents_get: Get document details and processing state. Guaranteed: accepts pofid or documents_id and returns identity, workspace and s
  • Po Documents Text Get: READ-ONLY. Returns data, changes nothing. po_documents_text_get: Retrieve page-level OCR/fulltext that already exists for a document. Guaranteed: returns only visual OCR text fr

Какие данные видит

Нужен ли аккаунт

Не нужен: сервер работает без входа

The AI document layer for your agent — 343 tools for business documents, EU-hosted.

  • Search & understand: full-text, semantic and hybrid search, document Q&A, knowledge graph
  • Capture & extract: OCR, IDP field extraction (invoices, contracts, delivery notes), classification
  • Process: PDF convert/merge/split/redact, anonymization (GDPR), URL-to-PDF, text-to-speech
  • Act: edit metadata, manage e-signature requests, workflows, audit trails, compliance (WORM)

Auth: OAuth 2.1 (recommended) or Authorization: Bearer from your PaperOffice account. Usage is credit-based — every result reports the credits charged. ISO 27001 & SOC 2 certified, hosted in Germany/Spain.

Docs: paperoffice.ai/en/developer/mcp · Examples: paperoffice-ai/cookbook

Как подключить

Как подключить PaperOffice Document Operations к Claude

Адрес сервера, который нужно вставить в настройках Claude:

https://document-operations--paperoffice.run.tools

  1. Откройте Claude (сайт claude.ai или приложение для компьютера).
  2. Перейдите в Настройки → Коннекторы (в английском интерфейсе: Settings → Connectors).
  3. Нажмите Добавить пользовательский коннектор (Add custom connector).
  4. Вставьте адрес сервера, скопированный ниже, в поле URL удалённого MCP-сервера и нажмите Добавить.
  5. В чате нажмите + → Коннекторы и включите новый коннектор.

Пользовательские коннекторы доступны на тарифах Free, Pro, Max, Team и Enterprise. На Free можно добавить только один. В Team и Enterprise коннектор сначала добавляет владелец организации в Настройки организации → Коннекторы.

Авторизация

После нажатия Добавить откроется окно входа в сервис. Войдите под своим аккаунтом и подтвердите доступ. Пароль Claude не видит.

Как подключить PaperOffice Document Operations к ChatGPT

Адрес сервера, который нужно вставить в настройках ChatGPT:

https://document-operations--paperoffice.run.tools

  1. Откройте ChatGPT в браузере (chatgpt.com). Нужен тариф Plus, Pro, Business, Enterprise или Edu.
  2. Один раз включите режим разработчика: Настройки → Приложения → Дополнительные настройки → Режим разработчика (Settings → Apps → Advanced settings → Developer mode).
  3. Откройте Настройки → Коннекторы и нажмите Создать (Create).
  4. Заполните форму: Название (любое), Описание (одна фраза, что делает сервис), URL MCP-сервера (скопируйте ниже).
  5. В поле Аутентификация выберите OAuth, если сервис требует вход, иначе Нет. Нажмите Создать.
  6. В новом чате откройте + → Приложения/Коннекторы и включите коннектор.

Название раздела у OpenAI менялось (Connectors → Apps/Plugins). Если пункт называется иначе, ищите «режим разработчика» в настройках. В рабочих пространствах Business/Enterprise администратор должен разрешить пользовательские коннекторы.

Авторизация

При первом использовании ChatGPT откроет окно входа в сервис. Войдите и подтвердите доступ.

Как подключить PaperOffice Document Operations к Cursor

Адрес сервера, который нужно вставить в настройках Cursor:

https://document-operations--paperoffice.run.tools

Быстрый способ: нажмите кнопку Открыть в Cursor ниже, Cursor сам предложит добавить сервер.

Вручную:

  1. В Cursor откройте Settings → Cursor Settings → MCP и нажмите Add new global MCP server.
  2. Вставьте JSON, скопированный ниже, в открывшийся файл ~/.cursor/mcp.json (для одного проекта: .cursor/mcp.json в корне проекта).
  3. Сохраните файл. Сервер появится в списке MCP; при необходимости пройдите авторизацию по кнопке рядом с ним.

Авторизация

Если сервис требует вход, рядом с сервером в списке MCP появится кнопка авторизации.

Список инструментов сервера (60)

Технические названия из tools/list. Нужны только разработчикам.

po_documents_searchREAD-ONLY. Returns data, changes nothing. po_documents_search: Stable document search for keyword, filename and available metadata. Guaranteed: validates required query/scope inputs, returns paginated document references with normalized core fields when available. Best effort: semantic/hybrid ranking, summaries, amounts and document_type depend on OCR/IDP/embedding coverage. Not for accounting totals. Score scale: relevance score is typically 0–100 (higher = better match) when sort_by=relevance (default for keyword search); use sort_by=modified_desc for recently changed documents. match_source values include exact, fulltext, ocr, entity, semantic. Browse/list all documents in a workspace: set query="*" or "**" with workspace_id (do NOT use po_documents_folders_list for that — folders_list returns only document folders, not files). Signature process filter: use signature_status (from signature_requests table) — NEVER fulltext-search for "signature request" in OCR. Keywords: search, find, query, list, documents, browse, enumerate, recently modified, signature status.
po_documents_getREAD-ONLY. Returns data, changes nothing. po_documents_get: Get document details and processing state. Guaranteed: accepts pofid or documents_id and returns identity, workspace and storage/processing metadata for accessible documents. Includes a compact trust block (score, verdict, finding counts, text_layer_ocr_mismatch) when a Trust certificate exists — read full findings with po_documents_trust_get, do not start po_documents_trust_agent_run just to inspect a result. Best effort: classification, financial fields, sender/recipient and OCR-derived metadata depend on AI-DMS/IDP coverage and may be absent. compact=true (schema default) returns core identity/classification/financial fields plus capabilities, processing_state and trust; compact=false additionally includes full meta_fields and extended document payload. Keywords: get document, details, metadata, info, trust certificate.
po_documents_text_getREAD-ONLY. Returns data, changes nothing. po_documents_text_get: Retrieve page-level OCR/fulltext that already exists for a document. Guaranteed: returns only visual OCR text from OCR pages or explicit OCR text fields; if no OCR exists it returns NO_OCR_TEXT instead of a summary. Poll po_documents_get (compact=true).processing_state until processing_status is processed, then retry this tool. It does not run OCR and never mixes hidden/injected PDF-layer text into the fulltext. When a Trust certificate exists, trust_text_signals.text_layer_ocr_mismatch flags a PDF-text-layer vs OCR divergence — use po_documents_trust_get for findings[].text_excerpt. Accepts pofid or documents_id. Default format is plaintext; format=html preserves sanitized inline OCR markup. Keywords: document text, read, content, OCR, fulltext, trust mismatch.
po_documents_trust_getREAD-ONLY. Returns data, changes nothing. po_documents_trust_get: Read the stored Trust Agent certificate for a document (score, verdict, findings with text_excerpt, tags, last run). This is a read — it does not start a scan and does not cost Trust-run credits. Use this before starting a new scan. For an adhoc scan use po_documents_trust_check (profile + tier, no agent). Use po_documents_trust_agent_run only for a saved Trust Agent. Hidden/injected PDF text is only in findings[].text_excerpt, never in po_documents_text_get.
po_documents_processing_agents_listREAD-ONLY. Returns data, changes nothing. po_documents_processing_agents_list: List document processing agents (pdf_trust, pdf_split, idp_extract, pdf_redact) with agent_id values for *_agent_run. Adhoc Trust uses po_documents_trust_check (no agent). Reading a Trust result does not need an agent id — use po_documents_trust_get.
po_documents_trust_checkWRITES DATA. Creates or changes data in the workspace. po_documents_trust_check: WRITE — start an adhoc Trust / integrity scan (credits). No trust_agent_id. Same engine as Documents-Overview quick check (the document_integrity_check workflow). Required: pofid or documents_id. Optional: profile (general|invoice|bank_statement|contract|identity_document|payslip|application|utility_bill|receipt|certificate|official_letter|medical_certificate|insurance|delivery_note|rental_agreement|power_of_attorney|vehicle_document|email_export), tier (basic|premium|ultra, default premium), include_visual_evidence, processing_lane (no_sla|sla_24h|sla_12h|sla_6h|sla_1h|instant — start SLA, credit factor). Ultra adds a semantic vision audit (semantic_audit block in the certificate). Poll the returned job_id with po_job_get until completed, then READ with po_documents_trust_get. Use po_documents_trust_agent_run only for a saved Trust Agent. Hidden/injected PDF text stays in findings[].text_excerpt.
po_entities_searchREAD-ONLY. Returns data, changes nothing. po_entities_search: Search available raw/canonical/meta-field-derived entities across processed documents. Guaranteed: validates query length and entity_type, returns referencable document-level rows when available. Best effort: entity coverage, page numbers, confidence and canonical source documents depend on AI-DMS/entity resolution coverage. Use workspace_id for scoped/sub-user searches. Keywords: search entities, find company, person, IBAN.
po_workspaces_listREAD-ONLY. Returns data, changes nothing. po_workspaces_list: List workspaces accessible to the current token. Returns workspace ID, name, workspace_tier, _capabilities (trash/recycle-bin, retention), documents_count and storage size. Primary tool for workspace discovery (paginate with limit/offset). Account-wide counts/statistics: use po_documents_stats_get (total_workspaces, by_workspace). Use workspace_id for a single lookup (minimal tokens). Default limit=25 to avoid LLM context overflow. documents_count = non-deleted rows incl. folders; stock_document_count = real documents only. Aliases: take=limit, skip=offset (clamped to 200). Keywords: list workspaces, workspace count, how many workspaces, folders, tier.
po_documents_vision_analyzeREAD-ONLY. Returns data, changes nothing. po_documents_vision_analyze: Get existing page preview images for visual analysis. Guaranteed: validates page syntax and returns available preview images. Best effort: images exist only when thumbnails/previews were generated; this tool does not perform vision extraction by itself. Accepts pofid or documents_id. Keywords: vision, analyze, image, visual, pages.
po_knowledge_graph_askREAD-ONLY. Returns data, changes nothing. po_knowledge_graph_ask: Ask best-effort cross-document questions using verified backend context plus Knowledge Graph/agent reasoning. Guaranteed: rejects too-short/off-domain questions and returns structured status/error fields. Best effort: answer quality depends on available KG, OCR, entities and financial read models; use po_analytics_financial_summary_get for deterministic totals (period=all for ranking). Keywords: knowledge graph, ask, query, GraphRAG, cross-document.
po_knowledge_graph_partners_getREAD-ONLY. Returns data, changes nothing. po_knowledge_graph_partners_get: Get best-effort business partner aggregates from the Knowledge Graph/read models. Guaranteed: returns partner rows available to the token and scope with explicit amount/date basis metadata. Best effort: canonicalization, amounts and last-document dates depend on entity resolution and financial meta-field coverage; not an accounting ledger. Keywords: partners, relationships, companies, persons, network.
po_intelligence_topics_getREAD-ONLY. Returns data, changes nothing. po_intelligence_topics_get: Get best-effort topics for one document (read-only, no side effects). Dedicated topics may be unavailable; when no topic pipeline data exists, MCP returns document_type as a pseudo-topic with source=document_type_fallback. Keywords: topics, categories, themes.
po_search_advancedREAD-ONLY. Returns data, changes nothing. po_search_advanced: Explicit document search modes. Guaranteed: validates query, workspace scope, mode and pagination and returns normalized document references. mode=exact performs strict post-filtering against the requested field when field_key is provided. semantic/hybrid are best effort and require embeddings. Score scale: MCP output uses normalized relevance scores 0–100 (score_scale=relevance_0_100) for all modes; raw cosine floats are not exposed. Debug/ranking internals are only intended when debug=true. Keywords: advanced search, semantic, fulltext, hybrid, exact.
po_documents_similar_findREAD-ONLY. Returns data, changes nothing. po_documents_similar_find: Find documents similar to a reference document using existing summary embeddings. Guaranteed: validates reference document and threshold/limit. Best effort: requires an embedding on the reference and comparable processed documents; low-quality type/file mismatches can be filtered from the MCP response. Keywords: similar, find, compare, duplicate, related.
po_documents_stats_getREAD-ONLY. Returns data, changes nothing. po_documents_stats_get: Get best-effort document statistics and coverage. Account scope (no workspace_id) returns metric_scope (acl_workspaces for group/limited tokens, full_account for admin tokens), total_workspaces, by_workspace[] and by_workspace_sum_matches_total. All headline totals respect the same workspace ACL as by_workspace. Counts are normalized for MCP clients and may reflect available OCR/IDP/entity/embedding tables, not guaranteed business totals. Use for overview and workspace count, not accounting. Keywords: statistics, stats, overview, count, workspace count, by_workspace, metric_scope, acl_workspaces, coverage.
po_entities_stats_getREAD-ONLY. Returns data, changes nothing. po_entities_stats_get: Get best-effort document statistics for one entity value. Guaranteed: returns matching document count/slice and aggregation scope. Best effort: amount totals are only meaningful for company-like document sets with extracted financial fields; documents can be truncated, so use returned count/scope metadata before interpreting aggregates. Keywords: entity, statistics, company stats, person, IBAN.
po_entities_canonical_getREAD-ONLY. Returns data, changes nothing. po_entities_canonical_get: List best-effort canonical entities from the entity resolution index. Canonicalization is useful for overviews, but it is not a guaranteed clean master-data registry. MCP filters IDP sentinels (VALUE_MISSING/VAL), empty canonical names, and invalid IBAN/phone/amount values. Keywords: canonical entities, deduplicated, list, companies, persons.
po_documents_thumbnail_getREAD-ONLY. Returns data, changes nothing. po_documents_thumbnail_get: Get an existing thumbnail/preview image for a specific document page. Guaranteed: validates document identity and page_number >= 1. Best effort: thumbnails exist only after preview generation/processing; missing thumbnails return a clear error. Accepts pofid or documents_id. Keywords: thumbnail, preview, image, page.
po_documents_upload_url_getWRITES DATA. Creates or changes data in the workspace. po_documents_upload_url_get: Registers an upload slot for a named file and returns its time-limited presigned HTTP PUT URL and upload_id. The result includes upload_url, optional curl_command, expires_in, and max_size_bytes. After the file is stored at upload_url, pass upload_id to po_documents_upload or another tool that accepts upload_id. If po_documents_upload is not in tools/list, execute it via po_mcp_tools_call_write with tool_id po_documents_upload. This tool does not accept file bytes or base64.
po_documents_uploadWRITES DATA. Creates or changes data in the workspace. po_documents_upload: Registers a new document in the PaperOffice DMS. Accepted file sources: upload_id from po_documents_upload_url_get after the file was stored at the returned upload_url; a public file_url for the server to fetch; or omit both to receive a browser upload page. File bytes and base64 are not accepted as tool arguments. Supports PDF, DOCX, XLSX, PNG, JPG, TIFF, and other common office formats. To request AI-DMS processing, set ai_dms_tier (basic/premium/ultra). When ai_dms_tier is set, processing_mode is sent as ai_dms and the backend queues or marks the document for processing; check po_documents_get.processing_state afterwards. Optional processing_lane sets Start-SLA (wait-to-start guarantee, not completion): no_sla (fair use ×1), sla_24h (×1.5), sla_12h (×2), sla_6h (×3), sla_1h (×4), instant (×5). If processing_lane is omitted, workspace default_processing_lane applies (else fair-use no_sla). Prefer processing_lane over legacy numeric priority. To rename the document, set file_name WITH extension (e.g. "my-report.pdf"). To add a NEW VERSION of an existing document, prefer po_documents_versions_create (or set version_of_pofid here). Keywords: upload, import, add document, processing_lane, start-sla, instant, new version.
po_documents_versions_createWRITES DATA. Creates or changes data in the workspace. po_documents_versions_create: Upload a changed file as the NEXT VERSION of an existing document. Pass version_of_pofid (POFID of any version of that document) plus the file via upload_id (from po_documents_upload_url_get) or file_url. The new version gets its own pofid and version number n+1; the previous version stays readable under its old pofid and the document keeps one entry in lists/search (latest only). Rules enforced by the API: same workspace, changed content (409 VERSION_SAME_CONTENT), version_comment required when the predecessor is revision-secure (WORM). Versioning is an AI-DMS feature: the source document must have an AI-DMS tier (basic, premium or ultra), otherwise the call fails with 402 VERSIONING_REQUIRES_AI_DMS - activate AI-DMS post-processing for that document first. The AI-DMS tier is inherited from the predecessor; a HIGHER tier may be requested via ai_dms_tier (it becomes the tier of the line), a lower tier is rejected with 400 VERSION_TIER_DOWNGRADE and processing_mode=no_ai_dms is rejected. Cost: 50 credits per created version (flat, includes the AI-DMS change summary) plus the upload volume and the AI-DMS page processing of the inherited tier. Keywords: new version, revision, replace file, update document, versioning, Version anlegen.
po_documents_versions_listREAD-ONLY. Returns data, changes nothing. po_documents_versions_list: Version history of a document line. Pass the pofid of ANY version; returns all versions (number, pofid, is_latest, comment, AI summary, creator, timestamps, protection by retention/legal hold, can_delete) plus latest_pofid. Default excludes versions in the recycle bin (include_trashed=true shows them). Each version is its own immutable pofid; use po_documents_get/po_documents_download with a version pofid to read that exact state. Keywords: versions, history, revisions, Versionshistorie, Stand.
po_documents_versions_restoreWRITES DATA. Creates or changes data in the workspace. po_documents_versions_restore: Rollback without rewriting history — takes an OLDER version (pofid) and creates a NEW latest version with that older content (n+1, own pofid). No IDP re-run: OCR text and metadata of the older version are copied. Fails if the selected version is already the current one or has the same content as the current one. Keywords: rollback, restore version, revert, Wiederherstellen als neue Version. Cost: 50 credits (same flat rate as creating a version); no upload volume and no AI-DMS page processing, because OCR and metadata are copied.
po_documents_versions_deleteDESTRUCTIVE. Changes or removes data, or reaches people outside this system (delete, revoke, cancel, send, activate retention). Not automatically undoable — confirm intent first. po_documents_versions_delete: Delete ONE older version (pofid) of a document line. The current (latest) version cannot be deleted on its own — delete the whole document with po_documents_delete (all versions, atomic: any retention/legal-hold protected version blocks the whole line) or create a newer version instead. Versions under retention (WORM) or legal hold cannot be deleted. mode=trash (default, restorable via po_documents_restore) or instant (permanent, confirm=true required). Keywords: delete version, remove revision, Stand loeschen.
po_documents_versions_diff_getREAD-ONLY. Returns data, changes nothing. po_documents_versions_diff_get: Compare a version with its predecessor (default) or with compare_to_pofid (same document line). Returns metadata deltas (size, pages, type, content_identical), OCR-text statistics (word-level similarity, added/removed words, changed pages) and the AI change summary when available. Text statistics are best effort and depend on OCR coverage. Keywords: compare versions, diff, what changed, Versionsvergleich.
po_documents_editDESTRUCTIVE. Changes or removes data, or reaches people outside this system (delete, revoke, cancel, send, activate retention). Not automatically undoable — confirm intent first. po_documents_edit: Edit document metadata fields. ALWAYS prefer short aliases as field keys — they are reliable and auto-resolve to internal paths. Common aliases: document_type (e.g. "Rechnung","Vertrag","Angebot"), sender_name, recipient_name, invoice_number, document_number, document_date, due_date, total_amount, net_amount, vat_amount, currency, locale, language, summary, keywords, workflow_state, document_status, save_paths. Example: {"fields": {"document_type": "Rechnung", "sender_name": "Firma GmbH", "total_amount": "1250.00"}}. You can also use full internal keys (with __ prefix) from po_documents_get, but short aliases are preferred for reliability. NOTE: Some internal keys use inconsistent _ vs __ prefixes (e.g. sender_name resolves to _document__data__parties__publisher_name). Fields are overwritten individually (other fields unchanged). Use po_documents_get first to see existing fields. Metadata editing does NOT require checkout, even in revision-secure workspaces (only file content changes require checkout). Non-numeric values in financial fields (total_amount, net_amount, vat_amount) trigger _type_warnings in the response. Non-numeric values in financial fields are rejected (not persisted).
po_documents_deleteDESTRUCTIVE. Changes or removes data, or reaches people outside this system (delete, revoke, cancel, send, activate retention). Not automatically undoable — confirm intent first. po_documents_delete: Delete one document (pofid or documents_id) or a short list of POFIDs; confirm=true is required above 5 POFIDs, and larger batches belong in po_documents_bulk_update with operation=delete. On business/compliance workspaces the document moves to the recycle bin and can be restored with po_documents_restore; standard-tier workspaces have no recycle bin, so deletion there is permanent. WORM/legal-hold documents may be skipped. Deleting is a SEPARATE permission from writing: an access token that may create, edit and upload in a workspace can still be refused here (error GROUP_RESTRICTION with restriction_kind=group_delete_denied). That is not a read-only workspace — check read_only_token and effective_write_permissions on po_documents_get before concluding the connection cannot write. Keywords: delete, remove, trash.
po_documents_moveWRITES DATA. Creates or changes data in the workspace. po_documents_move: Move a document to a different workspace. POFID stays the same, only location changes. Accepts pofid or documents_id. Keywords: move, relocate, transfer, document.
po_documents_copyWRITES DATA. Creates or changes data in the workspace. po_documents_copy: Create a copy of a document in another workspace. The copy gets a new POFID. Accepts pofid or documents_id. Optionally rename the file. Keywords: copy, duplicate, clone, document.
po_documents_processWRITES DATA. Creates or changes data in the workspace. po_documents_process: Trigger AI-DMS processing on a document that is ALREADY stored in PaperOffice. Use this when a document was uploaded without AI-DMS, or when you want to re-process it with a different tier. The document is marked/queued for AI-DMS processing; OCR, classification, metadata, entities and embeddings depend on the selected tier and backend worker completion. Existing analysis may be refreshed by the backend, but completion is asynchronous; check po_documents_get.processing_state afterwards. Same-tier reprocess after a successful run requires confirm_reprocess=true (HTTP 409 AI_DMS_REPROCESS_CONFIRM_REQUIRED otherwise). Tier downgrade requires confirm_tier_downgrade=true. First process and upgrades need no extra confirmation. Optional processing_lane sets Start-SLA (wait-to-start): no_sla (×1), sla_24h (×1.5), sla_12h (×2), sla_6h (×3), sla_1h (×4), instant (×5). If processing_lane is omitted, workspace default_processing_lane applies (else fair-use no_sla). Prefer processing_lane over legacy numeric priority. Free plans are always best effort (lane capped to no_sla). AI-DMS is PREPAID at submit: without enough credits the order is refused (HTTP 402 INSUFFICIENT_CREDITS — no job, no queue, document unchanged); 409 ORDER_IN_FLIGHT/ORDER_DUPLICATE/ORDER_UNVERIFIED = wait; 409 ORDER_IN_PROGRESS = document is being processed, wait; 409 ORDER_NOT_PENDING = already processed, pass requeue/process again; 503 BILLING_UNAVAILABLE = retry later. Re-ordering the same tier for a document that is already paid and waiting is idempotent (200, dedup, 0 credits). IMPORTANT: This does NOT upload a file — the document must already exist. Use po_documents_upload for new files. Accepts pofid or documents_id. Keywords: process, analyze, AI-DMS, reprocess, confirm_reprocess, confirm_tier_downgrade, OCR, processing_lane, start-sla, instant.
po_documents_downloadREAD-ONLY. Returns data, changes nothing. po_documents_download: Download the stored document file. Returns a temporary download URL (10min valid, default) that may be requested repeatedly (HEAD and GET) until it expires. Compliance note: PDF downloads can include PaperOffice metadata/branding added by the backend and may not be byte-identical to the originally uploaded file. Accepts pofid or documents_id. WARNING: base64 mode returns the entire file as text — for files >100KB this will exceed context limits. Use url mode (default) for most cases. Keywords: download, file, get document.
po_workspaces_createWRITES DATA. Creates or changes data in the workspace. po_workspaces_create: Create a workspace with mandatory business type (UI parity — e.g. kunde, mandant, projekt). Supports tier, AI-DMS, default_processing_lane (Start-SLA default), IDP pipeline, access protection, contacts, BYOS storage, and document-type preset. When omitted, workspace_tier defaults to business, ai_dms_mode to basic, default_processing_lane to no_sla / Best Effort (UI create parity). workspace_tier=standard: permanent deletes; business/compliance: recycle bin. Keywords: create, workspace, type, tier, start-sla.
po_workspaces_updateWRITES DATA. Creates or changes data in the workspace. po_workspaces_update: Update workspace fields (name, business type, tier, AI-DMS, default_processing_lane / Start-SLA default, IDP pipeline, access protection, contacts, storage, document-type preset). Use po_workspaces_list for IDs. Keywords: update, workspace, type, tier, start-sla.
po_workspaces_share_createDESTRUCTIVE. Changes or removes data, or reaches people outside this system (delete, revoke, cancel, send, activate retention). Not automatically undoable — confirm intent first. po_workspaces_share_create: Create a workspace share (email and/or telephony channel). Same as UI Freigabe tab. Requires workspace_id and at least one channel. Keywords: share, workspace, external, telephony.
po_workspaces_shares_listREAD-ONLY. Returns data, changes nothing. po_workspaces_shares_list: List workspace shares created by the current token (owner view). Optional workspace_id and include_history. Keywords: shares, list, owner, freigabe.
po_workspaces_shares_shared_with_meREAD-ONLY. Returns data, changes nothing. po_workspaces_shares_shared_with_me: List workspaces shared with the current token (recipient inbox). Keywords: shared, inbox, recipient.
po_workspaces_share_updateWRITES DATA. Creates or changes data in the workspace. po_workspaces_share_update: Update share permissions and settings (owner only). access_level is read (view + download) or upload (+ own uploads); write/admin/view are rejected with INVALID_ACCESS_LEVEL. Keywords: share, update, permissions.
po_workspaces_share_revokeDESTRUCTIVE. Changes or removes data, or reaches people outside this system (delete, revoke, cancel, send, activate retention). Not automatically undoable — confirm intent first. po_workspaces_share_revoke: Revoke share as owner or leave as recipient. Requires confirm=true. Keywords: revoke, leave, share.
po_workspaces_share_acceptWRITES DATA. Creates or changes data in the workspace. po_workspaces_share_accept: Accept share invitation via share_token from email link. Keywords: accept, invite, share.
po_workspaces_share_accept_by_idWRITES DATA. Creates or changes data in the workspace. po_workspaces_share_accept_by_id: Accept pending share from inbox by share_id. Keywords: accept, inbox, share_id.
po_workspaces_share_declineWRITES DATA. Creates or changes data in the workspace. po_workspaces_share_decline: Decline share invitation by share_id. Keywords: decline, reject, share.
po_workspaces_share_resend_inviteDESTRUCTIVE. Changes or removes data, or reaches people outside this system (delete, revoke, cancel, send, activate retention). Not automatically undoable — confirm intent first. po_workspaces_share_resend_invite: Resend pending share invitation email (owner, rate-limited). Keywords: resend, invite, share.
po_workspaces_share_cancel_inviteDESTRUCTIVE. Changes or removes data, or reaches people outside this system (delete, revoke, cancel, send, activate retention). Not automatically undoable — confirm intent first. po_workspaces_share_cancel_invite: Cancel pending invitation (owner). Requires confirm=true. Keywords: cancel, invite, share.
po_workspaces_share_infoREAD-ONLY. Returns data, changes nothing. po_workspaces_share_info: Public share metadata for signup/onboarding (token-based, bearer optional). Keywords: share, info, token, signup.
po_workspaces_external_users_listREAD-ONLY. Returns data, changes nothing. po_workspaces_external_users_list: List external users with workspace access (admin view, root/administrator). Keywords: external, users, admin, sharing.
po_workspaces_security_check_accessREAD-ONLY. Returns data, changes nothing. po_workspaces_security_check_access: Check whether a workspace requires password/SMS unlock and whether the current bearer token already has a valid workspace session. Returns protection_type (none|password|sms|password+sms), masked phone numbers, dual-auth flag, ACL membership (has_access), and active_session expiry. Call this before po_documents_* on protected/compliance workspaces. Keywords: 2FA, SMS, unlock, compliance, workspace security, access check.
po_documents_folders_createWRITES DATA. Creates or changes data in the workspace. po_documents_folders_create: Group multiple documents into a document folder/staple. Minimum 2 POFIDs required. Use to group invoice + delivery note + payment, bundle contract + attachments. Keywords: folder, staple, group, bundle.
po_documents_folders_dissolveDESTRUCTIVE. Changes or removes data, or reaches people outside this system (delete, revoke, cancel, send, activate retention). Not automatically undoable — confirm intent first. po_documents_folders_dissolve: Dissolve/ungroup a document folder — individual documents are preserved, only the grouping is removed. Keywords: dissolve, ungroup, unstaple, folder.
po_signing_request_createDESTRUCTIVE. Changes or removes data, or reaches people outside this system (delete, revoke, cancel, send, activate retention). Not automatically undoable — confirm intent first. po_signing_request_create: Create a digital signing request for a document and send invitations to signers. IMPORTANT: Every signer MUST have both email AND phone number. Phone is MANDATORY for SMS-OTP security verification — without a phone number the signing process WILL FAIL! SMS is only used for the one-time code: no SMS is sent at create — the signer receives the code on the signing page and signs there. This connector never signs on behalf of a signer. channel=email sends the invitation email; channel=manual or none returns the signing URL only, for the user to pass on. Always ask the user for the signer phone number before calling this tool. CRITICAL: The phone must be in full E.164 format with the REAL country code — NEVER auto-guess, assume or fabricate a country prefix like +49. If the user provides a number without + or country code, ask the user for the country code before calling this tool. A wrong country code causes SMS delivery to a wrong number or country. Supports eIDAS-compliant signatures (SES, AES, QES), multi-step signing order, automatic reminders. Optional placement_template_uuid applies a saved per-page signature layout (created in Mobile In-Person Sign). Explicit signature_fields win if both are sent. last/all anchors resolve against document_page_count. Keywords: sign, create signature, signing request, digital signature, eIDAS.
po_signing_requests_listREAD-ONLY. Returns data, changes nothing. po_signing_requests_list: WHEN TO USE: signature status for specific documents (batch lookup by documents_id/pofid) or the detail of one request. For a plain status-filtered list of all signature requests use po_esignatures_list_signature_requests instead. Keywords: list, signing requests, status, signatures.
po_signing_request_manageDESTRUCTIVE. Changes or removes data, or reaches people outside this system (delete, revoke, cancel, send, activate retention). Not automatically undoable — confirm intent first. po_signing_request_manage: Manage an existing signing request when you do NOT know which mechanism created it — cancel, send reminder, or resend invitation. WHEN TO USE: this tool tries the e-signature request API first and falls back to the signing-link API, so it works for both. Prefer the dedicated tool when you know the origin: po_esignatures_cancel_signature_request / po_esignatures_send_reminder for requests created with po_esignatures_create_signature_request, and po_signature_links_cancel_signing_request / po_signature_links_resend_invitation for links created with po_signature_links_send_signing_request. Keywords: manage, cancel, remind, resend, signing request, unknown origin.
po_signing_document_downloadREAD-ONLY. Returns data, changes nothing. po_signing_document_download: Download a signed PDF document, the original, or the signature certificate/audit trail. Keywords: download, signed, PDF, certificate.
po_signing_verifyREAD-ONLY. Returns data, changes nothing. po_signing_verify: Verify, check and validate digital signatures on a document. Validates signature integrity, certificate chain, eIDAS compliance (SES/AES/QES). Accepts POFID, upload_id, or file_url. Keywords: verify, check, validate, signature.
po_signing_types_listREAD-ONLY. Returns data, changes nothing. po_signing_types_list: List available signature types and saved signature templates. Returns eIDAS signature levels (SES simple, AES advanced, QES qualified) and stored signature/initial templates for the account. Use before creating signing requests to know available options. Keywords: signature types, templates, eIDAS, SES, AES, QES, list signatures.
po_signing_template_saveWRITES DATA. Creates or changes data in the workspace. po_signing_template_save: Save a signature or initials template as base64 image for repeated use. Keywords: save, signature, template, initials.
po_signing_reminders_configureWRITES DATA. Creates or changes data in the workspace. po_signing_reminders_configure: Save (action=save) or read (action=get) the global reminder settings for signing requests. Configure when and how often reminders are sent. Keywords: configure, reminders, signing, settings.
po_signing_mobile_link_generateWRITES DATA. Creates or changes data in the workspace. po_signing_mobile_link_generate: Generate a QR-code-capable link for mobile document signing. Keywords: mobile, QR code, signing link.
po_documents_ai_splitWRITES DATA. Creates or changes data in the workspace. po_documents_ai_split: Automatically split a multi-page PDF into individual documents using AI Vision. Detects document boundaries in scanned stacks, inbox, archive digitization. Returns download URLs for split documents. File input: upload_id (preferred), file_url, or pofid. Call without file input to get an upload link. Keywords: split, PDF, AI, separate.
po_aiocr_analyzeWRITES DATA. Creates or changes data in the workspace. po_aiocr_analyze: AI-powered OCR document analysis — extract text, detect tables, layout, bounding boxes. Modes: text (basic + bounding boxes), grid (+ QR/barcode), complete (+ tables/layout/markdown). Optional: create searchable PDF. File input: upload_id (preferred), file_url, or pofid. Call without file input to get an upload link. Keywords: OCR, AI, analyze, extract text, tables.
po_anonymization_pii_detectREAD-ONLY. Returns data, changes nothing. po_anonymization_pii_detect: Step 1 of document anonymization — detect personally identifiable information (PII): names, addresses, phone, email, IBAN, tax IDs, dates. Returns bounding boxes for step 2 (po_anonymization_redaction_apply). File input: upload_id (preferred), file_url, or pofid. Pass scenario for policy (default gdpr_disclosure; e.g. external_sharing, financial_data). Call without file input to get an upload link. Keywords: PII, detect, anonymize, GDPR, scenario.