power-automate-docrouter

Microsoft Power Automate tooling for DocRouter.ai: the DocRouter Organization RC1 custom connector (organization-scoped HTTP API under /v0/orgs/...) and the DocRouter Account custom connector (account-scoped APIs under /v0/account/...: users, organizations, tokens, AWS and LLM configuration).

Sandbox sample flow

The sandbox is a minimal Power Automate flow that proves end-to-end file ingestion: read a file from SharePoint (or any source that exposes a name + Base64 content) and send it to DocRouter with Upload Document. It is not a separate product—just a reference pattern you can copy.

Typical sequence:

1. Manually trigger a flow — run on demand while you test.
2. Get file metadata using path — resolves the file in a library (name, path, identifiers); the screenshot uses SharePoint.
3. Get file content — loads the file bytes for the next step.
4. Upload Document (DocRouter Organization RC1) — maps File name → document_name and File content → content, then runs OCR/LLM processing on the DocRouter side.

The same connector exposes the full org API (documents, tags, prompts, schemas, knowledge bases, OCR/LLM helpers, webhooks, and more). The action picker shows everything available under DocRouter Organization RC1:

Installing the connector

You need the Power Platform CLI connector tools (paconn), a Python 3 environment, and rights to create custom connectors in your Power Platform environment.

1. Clone and Python environment

cd power-automate-docrouter
python -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install -r requirements.txt

This installs paconn.

2. Sign in to Power Platform

From the repo root:

./login.sh

Or: paconn login after activating the venv. Complete the browser/device login for the tenant where the connector should live.

To clear stored credentials: ./logout.sh (or paconn logout).

3. Create a connector (first time)

Organization connector (documents, prompts, org APIs):

./docrouter/docrouter-org/create.sh

paconn registers the connector using docrouter/docrouter-org/apiDefinition.swagger.json, apiProperties.json, icon.png, and script.csx (injects organization id into /v0/orgs/... paths).

Account connector (users, organizations, account tokens, AWS, LLM providers — no script):

./docrouter/docrouter-account/create.sh

Uses docrouter/docrouter-account/apiDefinition.swagger.json, apiProperties.json, and icon.png only.

Each create.sh run prints the new connector id on stdout and suggests ORG_CONNECTOR_ID / ACCOUNT_CONNECTOR_ID for your repo-root .env (it also writes settings.json in that connector folder via paconn --overwrite-settings). You can also copy the id from Data → Custom connectors in Power Automate.

4. Update an existing connector

Set the connector id, then run update.sh for the connector you are refreshing:

export ORG_CONNECTOR_ID='shared_your-organization-connector-id-here'
export ACCOUNT_CONNECTOR_ID='shared_your-account-connector-id-here'
# or add ORG_CONNECTOR_ID / ACCOUNT_CONNECTOR_ID to a repo-root `.env` file
./docrouter/docrouter-org/update.sh
# or
./docrouter/docrouter-account/update.sh

You can also pass the id as the first argument:

./docrouter/docrouter-org/update.sh 'shared_your-organization-connector-id'
./docrouter/docrouter-account/update.sh 'shared_your-account-connector-id'

Optional: use a per-connector settings.json under docrouter/docrouter-org/ or docrouter/docrouter-account/ with paconn update --settings ... if you rely on saved paths (see Microsoft’s paconn documentation).

5. Use the connectors in a flow

In Power Automate, add an action from DocRouter Organization RC1 and create a connection when prompted:

Field	Purpose
Base URL	Optional. Default production API is used if blank (https://app.docrouter.ai/fastapi).
Organization ID	Your org segment from the DocRouter app URL (.../orgs/{organization_id}/...) or Settings → Organization.
Organization Token	API key from DocRouter Settings → Developer (sent as X-Api-Key).

After the connection succeeds, you can build flows such as the sandbox above or call Get Document, List Documents, tags, prompts, and the rest of the operations shown in the action list screenshot.

DocRouter Account connection

When you add an action from DocRouter Account, create a connection with:

Field	Purpose
Base URL	Optional. Default production API if blank (https://app.docrouter.ai/fastapi).
Account API token	Account-level token (acc_...) from DocRouter Settings → Developer. Organization tokens (org_...) do not work for /v0/account/ APIs.

Alternative: import in the portal

You can also create or refresh a custom connector manually in Power Automate or the Power Platform portal by importing the same OpenAPI (apiDefinition.swagger.json), properties (apiProperties.json), icon, and (for the organization connector only) the C# script—paconn is the supported path for keeping this repo and the tenant in sync.

Operations and parameters

Organization connector

The tables below mirror docrouter/docrouter-org/apiDefinition.swagger.json. Every HTTP operation lives under /v0/orgs/{organization_id}/.... The organization_id path segment is not filled in per action in the designer: it is injected from the connection (see script.csx and apiProperties.json). Query and path parameters marked internal or advanced in the spec may be hidden or collapsed in the UI.

In Body rows, a trailing * on a property name (for example name*) means required in the JSON body.

Documents

Summary	Operation ID	HTTP	Parameters (excluding organization_id)
Delete Document	DeleteDocument	DELETE	document_id (path, string, req)
Get Document	GetDocument	GET	document_id (path, string, req) include_content (query, boolean, opt, advanced)
List Documents	ListDocuments	GET	skip (query, integer, opt, advanced) limit (query, integer, opt, advanced) tag_ids (query, string, opt, advanced) name_search (query, string, opt, advanced)
Update Document Metadata	UpdateDocument	PUT	document_id (path, string, req) Body (DocumentUpdate): document_name, metadata (JSON string), tag_ids
Upload Document	UploadDocument	POST	Body (DocumentUpload): content* (string, byte / Base64), document_name*, metadata (JSON string), tag_ids

Knowledge Bases

Summary	Operation ID	HTTP	Parameters (excluding organization_id)
Create Knowledge Base	CreateKnowledgeBase	POST	Body (KnowledgeBaseConfig): name*; description, embedding_model, system_prompt, tag_ids
Delete Knowledge Base	DeleteKnowledgeBase	DELETE	kb_id (path, string, req)
Get Knowledge Base	GetKnowledgeBase	GET	kb_id (path, string, req)
List Knowledge Bases	ListKnowledgeBases	GET	name_search, skip, limit (query, opt, advanced)
Reconcile Knowledge Base	ReconcileKnowledgeBase	POST	kb_id (path, string, req)
Search Knowledge Base	SearchKnowledgeBase	POST	kb_id (path, string, req) Body (KBSearchRequest): query*; metadata_filter, top_k
Update Knowledge Base	UpdateKnowledgeBase	PUT	kb_id (path, string, req) Body (KnowledgeBaseConfig): name*; description, embedding_model, system_prompt, tag_ids

LLM

Summary	Operation ID	HTTP	Parameters (excluding organization_id)
Delete LLM Result	DeleteLLMResult	DELETE	document_id (path, string, req) prompt_revid (query, string, req)
Download LLM Result as JSON	DownloadLLMResult	GET	document_id (path, string, req)
Get LLM Extraction Result	GetLLMResult	GET	document_id (path, string, req) prompt_revid (query, string, opt, advanced)
List Available LLM Models	ListOrgLLMModels	GET	chat_only (query, boolean, opt, advanced)
Run LLM Extraction on Document	RunLLM	POST	document_id (path, string, req) prompt_revid, force (query, opt, advanced)
Update and Verify LLM Result	UpdateLLMResult	PUT	document_id (path, string, req) prompt_revid (query, string, req) Body (UpdateLLMResultRequest): updated_llm_result, is_verified

OCR

Summary	Operation ID	HTTP	Parameters (excluding organization_id)
Export OCR as HTML	ExportOCRHtml	GET	document_id (path, string, req)
Export OCR as Markdown	ExportOCRMarkdown	GET	document_id (path, string, req)
Get OCR Metadata	GetOCRMetadata	GET	document_id (path, string, req)
Get OCR Text	GetOCRText	GET	document_id (path, string, req) page_num (query, integer, opt, advanced)
Run OCR on Document	RunOCR	POST	document_id (path, string, req) force, ocr_only (query, opt, advanced)

Prompts

Summary	Operation ID	HTTP	Parameters (excluding organization_id)
Create Prompt	CreatePrompt	POST	Body (PromptConfig): name, content; model, schema_id, tag_ids, kb_id
Delete Prompt (by Prompt ID)	DeletePrompt	DELETE	prompt_id (path, string, req)
Get Prompt by Revision ID	GetPromptByRevId	GET	prompt_id (path, string, req)
List Prompt Versions	ListPromptVersions	GET	prompt_id (path, string, req)
List Prompts	ListPrompts	GET	name_search, skip, limit (query, opt, advanced)
Update Prompt (by Prompt ID)	UpdatePrompt	PUT	prompt_id (path, string, req) Body (PromptConfig): name, content; model, schema_id, tag_ids, kb_id

Schemas

Summary	Operation ID	HTTP	Parameters (excluding organization_id)
Create Schema	CreateSchema	POST	Body (SchemaConfig): name, response_format (JSON Schema object)
Delete Schema (by Schema ID)	DeleteSchema	DELETE	schema_id (path, string, req)
Get Schema by Revision ID	GetSchemaByRevId	GET	schema_id (path, string, req)
List Schema Versions	ListSchemaVersions	GET	schema_id (path, string, req)
List Schemas	ListSchemas	GET	name_search, skip, limit (query, opt, advanced)
Update Schema (by Schema ID)	UpdateSchema	PUT	schema_id (path, string, req) Body (SchemaConfig): name, response_format
Validate Data Against Schema	ValidateSchema	POST	schema_id (path, string, req) Body (SchemaValidateRequest): data*

Tags

Summary	Operation ID	HTTP	Parameters (excluding organization_id)
Create Tag	CreateTag	POST	Body (TagConfig): name*; color, description
Delete Tag	DeleteTag	DELETE	tag_id (path, string, req)
Get Tag	GetTag	GET	tag_id (path, string, req)
List Tags	ListTags	GET	name_search, skip, limit (query, opt, advanced)
Update Tag	UpdateTag	PUT	tag_id (path, string, req) Body (TagConfig): name*; color, description

Triggers

Summary	Operation ID	HTTP	Parameters (excluding organization_id)
When a DocRouter Event Occurs	OnDocRouterEvent	POST	Body: callbackUrl*; events (array)

Account connector

See docrouter/docrouter-account/apiDefinition.swagger.json for the full list. Summary:

Area	Operations (examples)
Organizations	List, create, update, delete (/v0/account/organizations)
Users	List, create, update, delete (/v0/account/users)
API tokens	List, create, delete account tokens (/v0/account/access_tokens)
AWS	Get, set, delete AWS config (/v0/account/aws_config) — system admin only
LLM	List models, list providers, set provider config (/v0/account/llm/...) — admin for provider listing/config
Utilities	Resolve token to organization (ResolveTokenOrganization)

Regenerate the OpenAPI from route summaries when the backend changes: python3 docrouter/docrouter-account/generate_swagger.py.

Repository layout

• docrouter/docrouter-org/ — organization connector (apiDefinition.swagger.json, apiProperties.json, script.csx, create.sh, update.sh, validate.sh, download.sh).
• docrouter/docrouter-account/ — account connector (same scripts except no script.csx; generate_swagger.py builds apiDefinition.swagger.json).
• CLAUDE.md — detailed conventions for maintaining the connectors and syncing with the DocRouter API.