# Glean MCP

> Glean MCP searches your entire company's knowledge base, connecting conversations across every SaaS app—Jira, Confluence, Slack, etc. Instead of digging through dozens of tabs and siloed databases, you ask your agent a question once and get AI-generated answers synthesized from corporate data instantly.

## Overview
- **Category:** brain-trust
- **Price:** Free
- **Tags:** enterprise-search, knowledge-base, ai-assistant, saas-integration, information-retrieval, natural-language-processing

## Description

Getting accurate information used to mean logging into half a dozen different systems just to piece together the answer. Now, you can talk to your company's entire knowledge repository using natural conversation. This MCP connects directly to your enterprise account, giving your agent full visibility across all your connected applications—from project management logs to internal documents and employee profiles.

You don't need to remember where a piece of information lives. You just ask the question: 'What was our policy on Q3 marketing spend?' The MCP retrieves data from multiple sources, synthesizes an answer using advanced AI, and gives you one pure response. If you’re building your agent logic on Vinkius, this connector allows your agent to discover knowledge across apps like never before. It handles everything from finding a specific document ID to identifying the colleague who owns that policy.

## Tools

### get_answer
Retrieves specific billing information by running structured rules against active account data.

### autocomplete
Predicts potential page destinations using partial text inputs, helping you navigate large internal sites quickly.

### chat_completion
Manages ongoing text conversations while keeping track of the entire chat history for complex reasoning checks.

### custom_request
Executes specific, custom data requests across different internal systems using defined arrays.

### delete_document
Removes indexed documents permanently from the search index and blocks future retrieval of that content.

### search_datasource
Performs structural data extraction to query properties across different connected accounts or systems.

### index_document
Adds new documents, articles, or records into the corporate knowledge base for AI search access.

### search_people
Generates a structured JSON payload that identifies and retrieves contact details for specific employees within the company.

### search_docs
Searches for bounded records inside the platform, finding documents across all connected services.

### get_suggestions
Extracts rich data flags and suggestions by analyzing validation rules within your existing knowledge base.

## Prompt Examples

**Prompt:** 
```
Search for 'Q2 hiring plan' in all apps
```

**Response:** 
```
Searching corporate knowledge... I found 3 documents across Google Drive and Confluence. The most relevant is 'Final Hiring Plan Q2' (ID: doc_abc). Would you like me to get an AI-generated summary of its contents?
```

**Prompt:** 
```
Who knows about 'React Native' in my company?
```

**Response:** 
```
Searching people... I found 2 colleagues with 'React Native' listed in their skills: Alice Smith (Mobile Lead) and Bob Johnson (Senior Engineer). Would you like their contact details or recent Slack activity?
```

**Prompt:** 
```
Get AI answer for: 'What is our expense policy for business travel?'
```

**Response:** 
```
Synthesizing answer... According to the Employee Handbook in Confluence, our business travel policy allows for $50/day in meal reimbursements and requires all flights to be booked via the Corporate Portal at least 2 weeks in advance.
```

## Capabilities

### Discovering Corporate Knowledge
The MCP searches all connected SaaS applications to retrieve and synthesize answers based on company documents, chat logs, and project records.

### Finding People & Skills
It looks up active directory information, matching user names and skills across the organization so you can find who knows what.

### Managing Conversations
The MCP maintains historical context for complex chats, allowing your agent to handle multi-step reasoning over time without losing track of the thread.

### Indexing and Updating Data
You can upload custom documentation or index existing files directly into the corporate search system.

### Cleaning Up Data
It provides mechanisms to permanently delete indexed documents, ensuring sensitive data is removed from future searches and retrievals.

## Use Cases

### Finding a policy that spans multiple departments
A new marketing hire needs to know the expense rules. Instead of checking HR's Confluence site, they ask their agent: 'What’s the travel budget for Q4?' The MCP uses its search_docs capability to query both finance and HR documentation, giving a consolidated answer.

### Tracking down the subject matter expert
A product manager needs input on a new feature. They ask: 'Who knows about React Native in our system?' The MCP uses search_people to return two colleagues' names and titles, saving them from guessing who to bother.

### Getting a summary of a large project
An executive needs the status of Project Phoenix. They ask their agent: 'What are the risks and next steps for Project Phoenix?' The MCP aggregates data by running structural extraction against multiple sources (Jira, Slack) to build a comprehensive report.

## Benefits

- Stop manual searching. Instead of jumping between Jira and Confluence to find one policy, you ask your agent a single question, and it returns an AI-synthesized answer pulling from all sources.
- Know who owns the knowledge. Use search_people to instantly identify colleagues with specific skills or roles, skipping weeks of internal email chains to find the right expert.
- Maintain context during complex tasks. The chat_completion tool keeps track of your entire conversation thread, so you don't have to restate background details when asking a follow-up question.
- Keep your data clean and secure. You can use delete_document to permanently remove indexed content, ensuring compliance and removing stale records from the search results.
- Build internal documentation dynamically. Indexing custom text properties allows developers to feed niche, proprietary knowledge directly into the corporate search engine for AI consumption.

## How It Works

The bottom line is you get instant access to your entire company's information, conversational style, without having to manually connect or query every single data source individually.

1. Subscribe to this MCP, then enter your specific Glean Domain name and API Token found in your Glean Admin Settings.
2. Pass these credentials to your preferred AI client, like Cursor or Claude, so it can access the corporate knowledge layer.
3. Ask your agent a question—like 'What was our Q2 hiring plan?'—and watch it pull synthesized answers from across all connected apps.

## Frequently Asked Questions

**How does Glean MCP search my company's data?**
The MCP connects to your enterprise account and uses its tools, like search_docs, to query multiple connected SaaS applications simultaneously. It synthesizes the results into one cohesive answer for you.

**Can Glean MCP find people by skill set?**
Yes. You can use the search_people tool to look up colleagues based on their listed skills and roles, giving you direct contact information or department details.

**Is this compatible with any AI agent?**
The MCP is designed to connect through Vinkius's standard Model Context Protocol (MCP). You just need your preferred AI client—Claude, Cursor, VS Code, etc.—to connect via the Vinkius platform.

**Does Glean MCP only read data?**
No. In addition to reading and searching, you can actively manage knowledge by using tools like index_document or delete_document to update your corporate search index.

**What if my company uses multiple versions of Confluence?**
The MCP is designed for enterprise-wide discovery. It handles the structural extraction necessary to pull accurate, relevant data across varied and complex internal knowledge sources.