Overview

The Guru MCP Server connects your Knowledge Agents to external AI systems using the Model Context Protocol (MCP).

This lets AI chat tools and agents ask questions, perform searches, and draft knowledge using the same governed company context that powers Guru’s own Answers, Chat, and Research.

MCP ensures these tools do not access raw documents or unverified data — they rely on Guru’s cited, permission-aware knowledge layer.

How it works

1. Connect once

Authenticate approved clients with OAuth, or use API tokens for broader agent frameworks and automation runtimes.

2. Expose your Knowledge Agents

External tools discover available Knowledge Agents and send questions, searches, and requests through them.

3. Retrieve governed answers

Every response includes:

• verified context
  
• permission-aware access
  
• citations to sources
  
• explainability and lineage
  

4. Draft & update content from anywhere

AI tools can create or update Guru content through MCP, triggering the same review and publishing workflows your teams use in Guru.

5. Governance stays intact

All MCP interactions flow into the AI Agent Center, where audit logs, lineage tracking, permissions, and verification workflows remain fully enforced.

Why it matters

Most enterprise AI pilots fail for one reason:
AI tools operate on stale, unverified, or permission-less data.

The Guru MCP Server solves this by giving every AI tool and agent access to the same governed knowledge layer, so answers are:

• Verified and permission-aware
  
• Cited and explainable
  
• Logged and auditable
  
• Aligned with internal policies and verification workflows

It’s interoperability without sacrificing accuracy, security, or trust.

Key capabilities

• Standards-based connectivity using MCP
  
• Knowledge Agent discovery & querying
  
• Cited, permission-aware answer retrieval
  
• Draft or update Guru Cards through MCP
  
• OAuth + API token authentication
  
• Full governance through the AI Agent Center
  
• Future-ready foundation for agentic workflows

Supported actions (Today)

AI chat tools and agents can:

• List Knowledge Agents
  
• Ask questions
  
• Retrieve cited, permission-aware answers
  
• Perform searches and Research
  
• Draft or update Guru content (where supported)
  
• Trigger human-in-the-loop publishing workflows
  

How it works

When connected, external AI tools can call the MCP Server to:

1. List available Knowledge Agents.
2. Ask questions or perform searches via those agents.
3. Retrieve answers that are verified, permission-aware, and cited.
4. Create or update Guru Cards from chat sessions.

OAuth connections are supported today for Claude and Cursor, with additional tools available upon request. For broader automation or iPaaS use cases, API token authentication is also supported.

Governance and control

Every MCP-powered interaction stays fully governed:

• Verified knowledge layer powers every answer
  
• Audit logs captured in the AI Agent Center
  
• Role-based access is enforced everywhere
  
• No duplication of permissions or RAG embeddings
  
• Single governed connection vs. many brittle pipelines

Outcomes

Connecting Guru through MCP isn’t just about interoperability—it’s about making every AI initiative succeed. Most failed AI projects suffer from the classic “garbage in, garbage out” problem. Guru MCP solves this by grounding external agents and workflows in verified company context. Instead of siloed, black-box experiments, every team and every AI agent benefits from the same AI Source of Truth—shared, governed, and continuously improving.

Here’s how it delivers that impact:

• Accelerate AI interoperability with ChatGPT, Claude, Cursor, Copilot, and more
  
• Reduce risk with governed, permission-aware context behind all answers
  
• Unlock automation with safe agentic workflows grounded in verified knowledge
  

Future-proof your stack with standards-based AI connectivity

Learn more about

• Knowledge Agents
• AI Agent Center‍
• Integrations‍
• Developer Network & APIs‍‍
• Research