Ragie
Fully managed RAG-as-a-Service for developers
Verdict
Common use cases
- Ingest product docs for AI-powered support
- Extract structured data from uploaded contracts
- Partition knowledge bases by client or project
- Query internal wikis during sprint planning
- Auto-process invoices with custom instructions
Integration
- Vendor
- Ragie
- Category
- developer-tools
- Auth
- API_KEY
- Tools
- 31
- Composio slug
ragie
Tools
- Create Document
Tool to upload and process a document file in Ragie. Use when you need to create a new document with support for various formats including text, images, and documents. The endpoint accepts multipart/form-data and returns a Document object w
- Create Document From URL
Tool to ingest a document from a publicly accessible URL. Use when you need to add documents to Ragie from external sources. The document undergoes processing steps (pending, partitioning, indexed, ready) before becoming available for retri
- Create Document Raw
Tool to ingest a document as raw text or JSON. Use when creating a new document from text or JSON data. The document goes through processing steps and becomes available for retrieval once in the ready state.
- Create Instruction
Tool to create a new instruction that applies natural language directives to documents as they're ingested or updated. Use when you need to define structured data extraction or analysis rules for documents in Ragie.
- Create OAuth Redirect URL
Tool to create an OAuth redirect URL for initializing embedded connector OAuth flows. Use when you need to set up OAuth authentication for connectors like Google Drive, Notion, or HubSpot.
- Create Partition
Tool to create a new partition for scoping documents and connections in Ragie. Use when you need to organize documents and set resource limits for different workspaces or tenants.
- Delete Documentdestructive
Tool to delete a document from Ragie. Use when you need to remove a document permanently from the system. Supports both synchronous and asynchronous deletion modes.
- Delete Instructiondestructive
Tool to delete an instruction and all associated entities. Use when you need to permanently remove an instruction (irreversible operation). Requires the instruction ID (UUID format).
- Delete Partitiondestructive
Tool to delete a partition and all associated data irreversibly. Use when you need to permanently remove a partition. Returns status 200 for synchronous deletion or 202 for asynchronous deletion.
- Get Document
Tool to retrieve a specific document by its unique identifier. Use when you need to get document details, metadata, processing status, or check for errors. Returns comprehensive document information including chunk count, page count, and an
- Get Document Chunk
Tool to retrieve a specific document chunk by its document and chunk ID. Use when you need detailed information about a specific chunk within a document, including its content, metadata, position index, and optional modality data for audio/
- Get Document Chunk Content
Tool to retrieve document chunk content in requested format with streaming support for media. Use when you need to get the actual content of a specific chunk from a document.
- Get Document Chunks
Tool to retrieve document chunks with pagination support. Lists all document chunks sorted by index in ascending order (max 100 items per page). Documents created prior to 9/18/2024 that have not been updated since have chunks sorted by ID
- Get Document Content
Tool to retrieve the content of a document by its ID. Use when you need to access the full content of a specific document. The media_type parameter can be used to request content in different formats.
- Get Document Summary
Tool to retrieve an LLM-generated summary of a document by its ID. Use when you need to get a concise summary of a document's content.
- Get Partition
Tool to retrieve a partition by ID with usage statistics and resource limits. Use when you need to get detailed information about a specific partition.
- Get Response
Tool to retrieve a response by its unique identifier. Use when you need to check the status or details of a previously created response.
- List Connections
Tool to list all connections sorted by creation date descending with pagination support. Use when you need to retrieve connections, optionally filtered by metadata.
- List Connection Source Types
Tool to list available connection source types like 'google_drive' and 'notion' along with their metadata. Use when you need to discover what connector types are available in Ragie.
- List Documents
Tool to list all documents sorted by creation date (descending) with pagination support. Use when you need to browse or retrieve document metadata. Supports filtering and pagination up to 100 items per page.
- List Entities By Document
Tool to retrieve all extracted entities from a specific document with pagination support. Use when you need to access structured data that has been extracted from a document by Ragie's entity extraction instructions.
- List Entities by Instruction
Tool to retrieve entities generated by a specific instruction. Use when you need to fetch entities extracted from documents based on a specific instruction's processing.
- List Instructions
Tool to retrieve all instruction records from the Ragie system. Use when you need to view all available instructions that define natural language prompts and entity schemas applied to documents.
- List Partitions
Tool to retrieve a paginated list of all partitions sorted by name in ascending order. Use when you need to list available partitions with their configurations and limits.
- Patch Document Metadata
Tool to update metadata for a specific document with partial update support. Use when you need to modify document metadata fields without replacing the entire metadata object. Supports both synchronous and asynchronous updates.
- Retrieve Document Chunks
Tool to retrieve relevant document chunks based on a query. Use when you need to search and retrieve document content that matches a specific query, with optional filtering and reranking capabilities.
- Set Partition Limits
Tool to set usage limits on partition pages and media. Use when you need to configure monthly or maximum limits for pages processed/hosted, video/audio processing, or media streaming/hosting for a specific partition.
- Update Document From URL
Tool to update an existing document by fetching content from a publicly accessible URL. Use when you need to refresh or replace a document's content with data from a web URL. The document goes through processing steps before it is ready for
- Update Document Raw
Tool to update a document's content from raw text or JSON data. Use when modifying existing document content. The document undergoes processing and becomes available for retrieval once it reaches the ready state.
- Update Instruction
Tool to update an instruction's active status. Use when you need to activate or deactivate an existing instruction.
- Update Partition
Tool to update a partition's configuration including description, context-aware settings, and metadata schema. Use when you need to modify an existing partition's settings.
Setup
Setup guide
- 11. Go to your Ragie dashboard and generate an API key under Settings or API Access. 2. In Switchy, open the Space where you want to use Ragie and click the integrations menu in the sidebar. 3. Select Ragie from the list of available MCPs and paste your API key into the authentication field. 4. Click Connect and wait for Switchy to verify the key (you'll see a green checkmark when it succeeds). 5. Test the connection by opening a chat in that Space and typing '@Ragie create a partition named test-workspace' — if Ragie responds with a partition ID, you're ready. 6. To ingest a document, use '@Ragie create document from URL' followed by a public link, or '@Ragie create document raw' with text content pasted inline. 7. Check document status by asking Ragie to list recent uploads; wait for 'ready' state before querying.
What teammates see: by default, memories from Ragie are scoped to the Space (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Spaces have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
Ingest Documentation Site
@Ragie create a document from this URL: https://docs.example.com/api-reference. Name it 'API Reference v2' and let me know when it's ready to query.Open in a Space →
Create Client Partition
@Ragie create a partition called 'Acme Corp Workspace' with a 500-document limit. I'll use this to scope all Acme-related files.Open in a Space →
Extract Invoice Fields
@Ragie create an instruction that extracts 'vendor_name', 'total_amount', and 'due_date' from any document tagged as invoice. Apply it to all future uploads in this partition.Open in a Space →
Upload Meeting Notes
@Ragie create a document from this raw text: [paste meeting notes here]. Title it 'Q1 Planning Meeting - Jan 15' and confirm when indexing finishes.Open in a Space →
Delete Outdated Content
@Ragie delete the document with ID abc-123-def. Run it asynchronously so it doesn't block other tasks.Open in a Space →
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@ragie search our product documentation for all mentions of API rate limits and return the top 5 most relevant chunks
Based on the indexed documents in your Ragie partition, here are the 5 most relevant passages about API rate limits: 1. From 'API Reference v2.1' (score: 0.94): Rate limits are set at 100 requests per minute for standard tier... 2. From 'Developer Guide' (score: 0.89): When you exceed rate limits, the API returns a 429 status code... 3. From 'Migration FAQ' (score: 0.82): Legacy rate limits of 50 req/min will sunset on March 2025... Each result includes the document ID, chunk text, and relevance score. This search queried across 847 indexed documents in your 'product-docs' partition.
This demonstrates Ragie's core retrieval capability — semantic search across your ingested documents. Requires documents to be uploaded and fully indexed (status: 'ready') first. Search quality depends on how you've partitioned your content and whether you've applied instructions for structured extraction. Results are scoped to the partition specified in your API key.
@ragie upload the file 'Q4-roadmap.pdf' from https://example.com/docs/Q4-roadmap.pdf and apply the 'extract-milestones' instruction to pull out all delivery dates
Document ingestion initiated. The PDF at https://example.com/docs/Q4-roadmap.pdf is now processing: - Document ID: doc_8x3k9m2p - Status: partitioning (will transition to indexed → ready) - Instruction applied: 'extract-milestones' will run once indexing completes Typically takes 30-90 seconds for a 10-page PDF. Once status reaches 'ready', the instruction will have extracted structured milestone data (dates, owners, deliverables) into the document's metadata, making it queryable via filtered retrieval.
This shows Ragie's document ingestion from URL plus instruction application in one step. The 'extract-milestones' instruction must already exist in your account. Document processing is asynchronous — the MCP returns immediately with a pending status. If the URL is inaccessible or the file format unsupported, processing fails silently unless you poll the document status. Instructions only execute after indexing completes, so there's a delay before structured data is available.
@ragie create a new partition called 'customer-support-tickets' with a 5000-document limit, then summarize how partitions help us isolate data for different teams
Partition created successfully: - Partition ID: part_7h2n4k8s - Name: customer-support-tickets - Document limit: 5,000 - Current document count: 0 Partitions in Ragie act as isolated namespaces for documents and connections. Each partition has its own resource limits and access scope, which is useful for multi-tenant applications or separating data by team/project. Documents in 'customer-support-tickets' won't appear in searches against other partitions unless explicitly queried. You can assign different API keys to different partitions to enforce access boundaries, and set per-partition limits to prevent runaway ingestion costs.
This example combines a write action (partition creation) with AI synthesis about the feature's purpose. Partitions are Ragie's primary data isolation mechanism — critical for SaaS apps serving multiple customers. Once created, you'll need to specify the partition ID when uploading documents or creating connections. Be aware that partition limits are hard caps; ingestion fails once you hit the document ceiling. Deleting a partition removes all its documents permanently.
Use-case deep-dives
When Ragie beats a static FAQ for support teams
A 6-person support team handling 200 tickets a week needs instant answers from product docs, past tickets, and Slack threads. Ragie wins here because its 31 tools let you ingest from URLs, raw text, and OAuth connectors (Google Drive, Notion) into one searchable partition. The instruction system means you can auto-tag documents as they come in—label a Notion page as 'billing policy' so retrieval prioritizes it for payment questions. The trade-off: if your knowledge base is under 50 documents and rarely changes, a simpler MCP like filesystem or a basic vector store is faster to set up. But once you're syncing multiple sources weekly and need scoped search per customer tier, Ragie's partition model justifies the API key overhead. Buy this if your support load is growing and you're tired of manually updating a wiki.
Ragie's partition model for per-customer doc isolation
A 12-person dev team building a B2B SaaS product needs each customer to upload their own compliance PDFs and internal runbooks, then query them without seeing other tenants' data. Ragie's Create Partition tool is purpose-built for this: each customer gets a partition with its own resource limits and document scope. You use Create Document From URL to pull in their Google Drive files via OAuth, then the instruction system extracts metadata (document type, department) as they're indexed. The boundary: if you're only serving 3-5 customers and documents are static, the 31-tool surface area is overkill—just use a folder-per-tenant setup in a simpler MCP. Ragie pays off at 20+ tenants or when customers upload documents weekly and expect instant semantic search. If you're scaling a multi-tenant AI feature, this MCP removes the partition-building work you'd otherwise write yourself.
When Ragie connects scattered sales collateral for deal prep
A 4-person sales team preps for calls by searching past proposals in Google Drive, deal notes in HubSpot, and pricing sheets in Notion. Ragie's OAuth connector tools (Create OAuth Redirect URL) let you sync all three sources into one retrieval layer, and the instruction system can auto-extract deal size or vertical from each document. During a call, the team queries 'enterprise pricing for healthcare' and gets ranked results across all sources. The catch: if your collateral lives in one place (just Google Drive) and you're not adding documents daily, a simpler filesystem or Google Drive MCP is faster. Ragie wins when you have 3+ disconnected sources and need cross-source search with metadata filtering. If your sales cycle depends on finding the right case study in under 10 seconds, the API key setup is worth it.
Frequently asked
What does the Ragie MCP do in Switchy?
It connects Switchy to your Ragie document retrieval system. Your team can upload documents, create partitions to organize content, and set instructions for how documents get processed. Useful when you're building AI workflows that need to search across your company's knowledge base without switching to the Ragie dashboard.
Do I need a Ragie API key to connect this MCP?
Yes. You'll need an API key from your Ragie account. Generate one in the Ragie dashboard under API settings, then paste it into Switchy's connection flow. The key grants access to all 31 tools, including document upload, partition management, and instruction creation. No OAuth flow required.
Can the Ragie MCP search or retrieve documents?
No, the MCP focuses on document management and setup. You can upload files, ingest from URLs, create partitions, and define instructions for processing. If you need to actually query or retrieve document content, you'll use Ragie's retrieval API separately or build that into your own workflow outside Switchy.
How is this different from using Ragie's dashboard directly?
The MCP lets your team manage Ragie documents inside Switchy's shared workspace, alongside other tools. You skip the context switch to the Ragie UI. Useful when you're chaining document ingestion with other automation steps. For one-off uploads or exploring Ragie's full feature set, the dashboard is still faster.
Who on the team should connect the Ragie MCP?
Whoever owns your Ragie account and can generate API keys. Typically a developer or ops lead. Once connected, any Switchy workspace member can use the tools to upload documents or create partitions. The API key's permissions apply to everyone, so don't share it with users who shouldn't delete documents.