Qoris MCP Server - Technical Reference

Complete technical documentation for integrating with the Qoris MCP server.

Server Information

MCP Server Name: Qoris
MCP Server URL: https://mcp.qoris.ai/mcp
Tagline: Governed AI agents with persistent memory, knowledge, MCP tools, and Knox safety Description: Qoris provides AI agents with persistent memory and knowledge management capabilities through Model Context Protocol. Store and retrieve information across conversations, search project knowledge bases, and build context-aware applications. Knox adds the governance layer for agents that execute commands, call tools, schedule work, or operate autonomously.

Server URL

Type: Custom MCP URLs (per user) Each user gets their own unique URL with their API key embedded as a query parameter: User URL Format: https://mcp.qoris.ai/mcp?api_key=YOUR_API_KEY Users can also use the Authorization header method with the base URL:

Base URL: https://mcp.qoris.ai/mcp
Header: Authorization: Bearer YOUR_API_KEY

How users get their URL:

Generate an API key in Qoris project settings
Copy the URL provided: https://mcp.qoris.ai/mcp?api_key=YOUR_API_KEY
Use this URL in their MCP client configuration

Connection Requirements

Setup Requirements:

Requires a Qoris account (free tier available)
Requires a project in your Qoris workspace
Requires an API key generated from project settings
No geographic restrictions
No admin or developer seat required

Getting Started:

Sign up at qoris.ai
Create a project
Generate an API key from project settings
Configure MCP connection with your API key

Authentication

Authentication Type: API Key or OAuth 2.0 Qoris supports three ways to authenticate MCP connections:

Method 1: OAuth 2.0 (For Cursor, Claude Code)

MCP clients that support OAuth use discovery endpoints (/.well-known/oauth-authorization-server, /.well-known/oauth-protected-resource), register dynamically (DCR), and redirect users to the consent page (https://app.qoris.ai/oauth/consent). Users sign in, select a workspace, approve, and the client receives an access token.

Access tokens do not expire — they remain valid until revoked
Revocation: Via Connected Apps in the Qoris dashboard, or DELETE /oauth/connections/<token_id>
Bearer method: Authorization: Bearer <access_token>

See Authentication for full OAuth flow details.

Method 2: API Key via Query Parameter (Simplest)

Pass the API key as a query parameter in the URL:

https://mcp.qoris.ai/mcp?api_key=YOUR_API_KEY

Convenient for quick testing and MCP clients that support URL-based auth.

Method 3: API Key via Authorization Header (Recommended for API Keys)

Pass the API key in the Authorization header:

{
  "headers": {
    "Authorization": "Bearer YOUR_API_KEY"
  }
}

More secure than query parameters and recommended for production. Note: The X-Qoris-Project-ID header is not required — the project ID is determined from the API key or OAuth token. API Key Generation:

Navigate to your Qoris project settings
Go to the “MCP” tab
Click “Create API Key” in the API Keys section
Copy and securely store the key (you won’t be able to see it again)

API keys provide full access to your project. Keep them secure and never commit them to version control.

Transport Support

Supported Transport: Streamable HTTP Qoris MCP server supports Streamable HTTP transport. SSE support may be added in the future, but Streamable HTTP is recommended for all integrations.

Read/Write Capabilities

Capability: Read + Write Qoris MCP server provides both read and write operations: Read Operations:

get_memories - Retrieve workspace memories
search_knowledge - Search project knowledge base
list_knowledge_documents - List available knowledge documents
get_document_full_content - Retrieve full document content

Write Operations:

save_memory - Create new memories
update_memory - Update existing memories
delete_memory - Delete memories

Knox Governance Layer

MCP controls how agents connect to Qoris memory and knowledge. Knox governs higher-risk actions around that access path. Use Knox when agents can:

execute shell commands or scripts
write or edit files
call MCP tools with untrusted external input
create scheduled or background tasks
invoke subagents
mutate external business systems
perform operations that require approval or audit

For developer-agent environments, see the Knox Claude Code Plugin. For the broader Qoris governance model, see the Knox Overview.

MCP App Status

Is this an MCP App? No Qoris MCP server is a standard MCP server without interactive UI elements. All interactions are through MCP tool calls.

Third-party Connections

Web Access: No
Third-party AI Model Integration: Yes (for knowledge search synthesis)
Third-party Data Retrieval: No
Third-party Data Modification: No

Qoris uses third-party AI models (via our backend) for knowledge search answer synthesis. The MCP server itself does not directly access external web services or third-party APIs beyond our own infrastructure.

Data Handling

Server only accesses data explicitly requested by user: Yes
No data is stored beyond session requirements: Yes (data persists per project, not per session)
Data transmission is encrypted (HTTPS/TLS): Yes
GDPR compliant: Yes (if applicable)

Data Storage:

Memories are stored in PostgreSQL databases
Knowledge items are stored in project-specific storage
All data is encrypted at rest and in transit
Data is scoped to projects and users

Data Retention:

Memories persist until explicitly deleted by the user
Knowledge items persist until removed from the project
API keys can be revoked at any time

Personal Health Data: No Qoris does not handle personal health data or medical records.

Use Cases & Examples

Use Case 1: Conversational AI with Memory

Scenario: Building a customer support chatbot that remembers user preferences and previous interactions. Example Prompt:

"Remember that this user prefers email notifications and has a premium subscription. 
Save this information for future conversations."

MCP Tool Call:

{
  "tool": "save_memory",
  "arguments": {
    "content": "User prefers email notifications and has premium subscription",
    "agent_id": "support_bot",
    "agent_name": "Customer Support Bot"
  }
}

Value: Enables personalized experiences across multiple conversations without requiring users to repeat information.

Use Case 2: Documentation Assistant

Scenario: Creating an AI assistant that can answer questions about your project’s documentation and codebase. Example Prompt:

"Search the knowledge base for information about authentication setup. 
How do I configure API keys?"

MCP Tool Call:

{
  "tool": "search_knowledge",
  "arguments": {
    "query": "How to configure authentication and API keys",
    "context": "Setting up a new project and need authentication guidance"
  }
}

Value: Provides instant access to project documentation, code examples, and knowledge without manual searching.

Use Case 3: Multi-Agent Workflow Coordination

Scenario: Multiple AI agents working on the same project need to share context and state. Example Prompt:

"Agent A completed the database migration. Save this status so Agent B knows 
the migration is done and can proceed with the next step."

MCP Tool Call:

{
  "tool": "save_memory",
  "arguments": {
    "content": "Database migration completed by Agent A at 2025-02-02 14:30 UTC",
    "agent_id": "agent_a",
    "agent_name": "Database Agent"
  }
}

Then Agent B retrieves:

{
  "tool": "get_memories",
  "arguments": {
    "limit": 10
  }
}

Value: Enables coordination between multiple agents working on complex workflows by sharing state and context.

Use Case 4: Code Review Assistant

Scenario: An AI assistant that reviews code and remembers common issues and patterns for your team. Example Prompt:

"Remember that our team always uses TypeScript strict mode and prefers async/await 
over promises. Save these coding standards."

MCP Tool Call:

{
  "tool": "save_memory",
  "arguments": {
    "content": "Team coding standards: Always use TypeScript strict mode, prefer async/await over promises",
    "agent_id": "code_reviewer",
    "agent_name": "Code Review Assistant"
  }
}

Value: Maintains consistent code quality by remembering team-specific standards and preferences.

Use Case 5: Project Knowledge Base Query

Scenario: Quickly finding information from uploaded documentation, meeting notes, or project files. Example Prompt:

"What are the deployment requirements mentioned in our project documentation? 
Search the knowledge base for deployment information."

MCP Tool Call:

{
  "tool": "search_knowledge",
  "arguments": {
    "query": "deployment requirements and procedures",
    "context": "Preparing for production deployment"
  }
}

Value: Instantly retrieves relevant information from all project documentation without manual file searching.

API Reference

Base URL

https://mcp.qoris.ai/mcp

Authentication

Option 1: OAuth 2.0

Authorization: Bearer <access_token>

Access tokens do not expire until revoked. See Authentication for OAuth flow. Option 2: API Key via Query Parameter

https://mcp.qoris.ai/mcp?api_key=YOUR_API_KEY

Option 3: API Key via Authorization Header

Authorization: Bearer YOUR_API_KEY

Available Tools

See individual tool documentation:

Error Handling

All tools return consistent error responses:

400 Bad Request: Invalid parameters or validation errors
401 Unauthorized: Invalid or missing API key
403 Forbidden: Permission denied (e.g., trying to update another API key’s memory)
404 Not Found: Resource not found (e.g., memory doesn’t exist)
500 Internal Server Error: Server-side errors

See individual tool documentation for specific error codes and messages.

Rate Limits

Rate limits are applied per API key:

Default: 1000 requests per hour
Premium: Higher limits available

Rate limit headers are included in responses:

X-RateLimit-Limit: 1000
X-RateLimit-Remaining: 999
X-RateLimit-Reset: 1640995200

Privacy & Security

Privacy Policy: https://qoris.ai/privacy Data Processing Agreement: Available upon request for enterprise customers Security:

All connections use HTTPS/TLS encryption
API keys are hashed and stored securely
Data is encrypted at rest
Regular security audits and updates

Support

Support Channel: [email protected] or GitHub Issues Documentation: https://docs.qoris.ai Status Page: https://status.qoris.ai

Configuration Examples

Cursor / Claude Code (OAuth)

MCP clients that support OAuth discover the server at https://mcp.qoris.ai/mcp and redirect users to the consent page. No manual configuration needed — users sign in and approve. Access tokens do not expire until revoked.

Claude Desktop (API Key)

Option 1: Using Authorization Header (Recommended) Add to claude_desktop_config.json:

{
  "mcpServers": {
    "qoris-memory": {
      "url": "https://mcp.qoris.ai/mcp",
      "transport": "streamable_http",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Option 2: Using Query Parameter

{
  "mcpServers": {
    "qoris-memory": {
      "url": "https://mcp.qoris.ai/mcp?api_key=YOUR_API_KEY",
      "transport": "streamable_http"
    }
  }
}

Custom MCP Client

import httpx

async def call_qoris_tool(tool_name, arguments, api_key):
    async with httpx.AsyncClient() as client:
        response = await client.post(
            "https://mcp.qoris.ai/mcp",
            headers={
                "Authorization": f"Bearer {api_key}",
                "Content-Type": "application/json"
            },
            json={
                "tool": tool_name,
                "arguments": arguments
            }
        )
        return response.json()

Versioning

Current Version: 1.0.0 API versioning is handled through the MCP protocol version. Breaking changes will be communicated through our changelog and documentation updates.

Changelog

See https://docs.qoris.ai/changelog for version history and updates.

Getting Started

MCP Tools

Knox

Technical Reference

Qoris MCP Server - Technical Reference

Server Information

Server URL

Connection Requirements

Authentication

Method 1: OAuth 2.0 (For Cursor, Claude Code)

Method 2: API Key via Query Parameter (Simplest)

Method 3: API Key via Authorization Header (Recommended for API Keys)

Transport Support

Read/Write Capabilities

Knox Governance Layer

MCP App Status

Third-party Connections

Data Handling

Categories

Use Cases & Examples

Use Case 1: Conversational AI with Memory

Use Case 2: Documentation Assistant

Use Case 3: Multi-Agent Workflow Coordination

Use Case 4: Code Review Assistant

Use Case 5: Project Knowledge Base Query

API Reference

Base URL

Authentication

Available Tools

Error Handling

Rate Limits

Privacy & Security

Support

Configuration Examples

Cursor / Claude Code (OAuth)

Claude Desktop (API Key)

Custom MCP Client

Versioning

Changelog

​Qoris MCP Server - Technical Reference

​Server Information

​Server URL

​Connection Requirements

​Authentication

​Method 1: OAuth 2.0 (For Cursor, Claude Code)

​Method 2: API Key via Query Parameter (Simplest)

​Method 3: API Key via Authorization Header (Recommended for API Keys)

​Transport Support

​Read/Write Capabilities

​Knox Governance Layer

​MCP App Status

​Third-party Connections

​Data Handling

​Categories

​Use Cases & Examples

​Use Case 1: Conversational AI with Memory

​Use Case 2: Documentation Assistant

​Use Case 3: Multi-Agent Workflow Coordination

​Use Case 4: Code Review Assistant

​Use Case 5: Project Knowledge Base Query

​API Reference

​Base URL

​Authentication

​Available Tools

​Error Handling

​Rate Limits

​Privacy & Security

​Support

​Configuration Examples

​Cursor / Claude Code (OAuth)

​Claude Desktop (API Key)

​Custom MCP Client

​Versioning

​Changelog

Qoris MCP Server - Technical Reference

Server Information

Server URL

Connection Requirements

Authentication

Method 1: OAuth 2.0 (For Cursor, Claude Code)

Method 2: API Key via Query Parameter (Simplest)

Method 3: API Key via Authorization Header (Recommended for API Keys)

Transport Support

Read/Write Capabilities

Knox Governance Layer

MCP App Status

Third-party Connections

Data Handling

Categories

Use Cases & Examples

Use Case 1: Conversational AI with Memory

Use Case 2: Documentation Assistant

Use Case 3: Multi-Agent Workflow Coordination

Use Case 4: Code Review Assistant

Use Case 5: Project Knowledge Base Query

API Reference

Base URL

Authentication

Available Tools

Error Handling

Rate Limits

Privacy & Security

Support

Configuration Examples

Cursor / Claude Code (OAuth)

Claude Desktop (API Key)

Custom MCP Client

Versioning

Changelog