# Glean MCP > Glean connects your AI agent to all your company's data—Confluence, Slack, Google Drive, Jira, and more. It lets you ask natural questions about any internal document or person, getting instant answers without clicking through dozens of siloed apps. ## Overview - **Category:** productivity - **Price:** Free - **Tags:** enterprise-search, unified-search, people-discovery, ai-qa, knowledge-base, data-indexing ## Description Imagine needing an answer that lives somewhere between the engineering wiki, a Slack thread from three months ago, and a payroll sheet in Google Drive. Usually, you'd spend 20 minutes opening five different tabs just to piece together the context. This MCP lets your AI client bypass all that friction. Instead of searching tool by tool, you ask one question—like 'Who owns the API documentation for the payments team?'—and get a single answer compiled from every connected source. You don't manage data sources; you just talk to them. By connecting Glean through Vinkius, your agent instantly gains visibility across your entire enterprise knowledge base, turning fragmented documents and chat logs into actionable intelligence. ## Tools ### autocomplete Provides intelligent search suggestions based on known company terminology as you type. ### bulk_index_documents Indexes a large number of documents to make them searchable at once. ### chat Asks natural language questions and gets AI-generated answers grounded in your internal knowledge base. ### check_glean_status Verifies the current connectivity status of the entire MCP to ensure data is flowing correctly. ### delete_document Permanently removes a specific document from your connected knowledge sources. ### get_collection Retrieves detailed information about a specific, curated set of related documents. ### get_document Gets the metadata and details for one single document file based on its ID. ### index_document Adds a single, new document to your searchable knowledge repository. ### list_collections Shows you all the pre-built and curated groups of documents available in your system. ### search_by_datasource Narrows a general search to look only within one specific connected source, like Confluence or Jira. ### search_people Finds employee records by name, department, role, or area of expertise across the organization. ### search Performs a broad search query that scans content across every connected platform simultaneously. ## Prompt Examples **Prompt:** ``` Search for our deployment runbook and the on-call rotation schedule. ``` **Response:** ``` Found 2 relevant results across your connected sources. 1) 'Production Deployment Runbook v3.2' — Confluence (Engineering wiki), last updated 3 days ago. Steps: pre-deploy checklist, rollback procedure, monitoring dashboards. 2) 'On-Call Rotation Q2 2025' — Google Sheets, shared by Sarah Chen. Current on-call: Mike Torres (Backend), Lisa Park (Frontend), expires May 2. Would you like to search for related incident reports or ask the AI assistant for a deployment summary? ``` **Prompt:** ``` Ask the AI assistant: What is our company's refund policy for enterprise customers? ``` **Response:** ``` Based on your organization's knowledge base, the Glean AI assistant responds: 'Enterprise customers are eligible for a full refund within 30 days of contract signing. After 30 days, refunds are prorated based on remaining contract months. Refund requests must be submitted through the Customer Success team via Zendesk. All enterprise refunds require VP-level approval. Source: Customer Policy Handbook (Confluence, updated Feb 2025).' Would you like to see the full policy document or search for recent refund requests in Zendesk? ``` **Prompt:** ``` Find the engineering lead for the payments team and search Slack for recent discussions about PCI compliance. ``` **Response:** ``` People search result: James Rodriguez (james@company.com) — Senior Engineering Manager, Payments Team. Reports to: VP Engineering. Located: San Francisco. Expertise: payment processing, PCI-DSS, Stripe integration. Slack search results for 'PCI compliance': 3 relevant threads in #payments-eng. 1) 'PCI audit prep checklist' — 2 days ago, 12 replies. 2) 'TLS 1.3 migration timeline' — 5 days ago, 8 replies. 3) 'Tokenization vendor review' — 1 week ago, 15 replies. Would you like to read a specific thread or ask the AI for a PCI compliance summary? ``` ## Capabilities ### Cross-Platform Search Search content simultaneously across dozens of connected apps like Confluence, Slack, and Jira. ### Knowledge Q&A Chat Ask natural language questions to generate answers that are directly cited from your company's own documentation. ### People Directory Lookup Find employee details, including their role, expertise, and department, across the entire organization. ### Targeted Search Filtering Focus your search results only on a single platform, like pulling everything from just Google Drive. ### Content Management Programmatically index new documents or manage curated knowledge collections for IT teams. ## Use Cases ### Troubleshooting a legacy API endpoint A developer needs to know the original owner and implementation notes for an old payment microservice. Instead of guessing which department owns the documentation, they use `search_people` to find the former lead engineer, then run a general `search` query focused on that person's name in Confluence. ### Preparing for an internal audit The compliance officer needs proof of all current security policies. They use `list_collections` to find the official 'Security Policy' collection, then run a specific search on that collection using `get_collection` details to compile the necessary reports. ### Onboarding a new hire A new team member can ask the AI agent: 'What are our current vacation policies?' The chat function responds with policy documents found across HR Drive and Confluence, all in one summarized response. ## Benefits - Find answers instantly. Instead of manually opening Jira, Confluence, and Drive to answer a question, using the `chat` tool lets you ask your agent once and get all relevant results cited from multiple sources. - Stop wasting time jumping between tabs. The general `search` function reads across every connected app—people records, wikis, chats—making discovery feel like one unified database. - Target specific knowledge sets. If you know the answer is only in Slack threads but don't want to wade through Jira tickets, use `search_by_datasource` to limit your query scope. - Manage content programmatically. For IT teams, functions like `index_document` and `bulk_index_documents` let you update knowledge sources without manual intervention. - Discover people faster than HR can find them. The dedicated `search_people` tool lets you look up expertise or department leads instantly, solving that 'who do I ask?' problem. ## How It Works The bottom line is that you treat your entire company's knowledge like one giant, searchable file cabinet. 1. Subscribe to this MCP and provide your Glean API Token in the Vinkius admin console. 2. Your agent uses the connection to pull data from all connected enterprise sources. 3. You ask a question or execute a search query through your AI client, receiving consolidated results. ## Frequently Asked Questions **How do I check if the Glean MCP is working correctly?** Run the `check_glean_status` tool. This instantly verifies connectivity and confirms that all your connected data sources are available for searching. **Can I use search_by_datasource to limit my query?** Yes. Using `search_by_datasource`, you tell the MCP to look only within a single platform, like Confluence or Google Drive, which is useful when you know where the answer must live. **How do I find an employee's expertise?** Use `search_people`. You can search by name, role, or specific area of expertise to locate who knows what within your organization. **What is the difference between general search and chat?** General `search` returns multiple snippets of text from where it found information. The `chat` tool uses AI to synthesize those results and give you a single, summarized answer. **When I use `bulk_index_documents`, how does that affect my existing knowledge base?** It processes and adds new content without disrupting live searches. You can upload large batches of documents to immediately expand your searchable data pool. **If I run `list_collections`, what kind of information do I receive about my content groups?** The tool returns a list of all curated collections associated with your workspace. From this list, you can select specific groupings to view their contents and policies. **What is the process when I use the `delete_document` tool?** This action permanently removes a specified document from the index and search results. Use it carefully, as this deletion cannot be undone once executed. **How does running `autocomplete` help me find information faster?** It provides intelligent suggestions based on the full scope of your connected knowledge. As you type, it anticipates relevant terms and people mentioned across all sources. **Can I search across multiple data sources like Confluence, Slack, and Google Drive simultaneously?** Yes! The `search` tool queries all connected data sources in a single request. If you want to narrow results to a specific platform, use `search_by_datasource` with the datasource name (e.g., 'Confluence', 'Slack', 'Google Drive'). Use `autocomplete` to get intelligent suggestions as you refine your query. **Can I ask the Glean AI assistant questions about my organization's knowledge?** Yes. The `chat` tool sends a natural language question to Glean's AI assistant, which generates an answer grounded in your organization's indexed knowledge — including documents, wikis, Slack conversations, and Jira tickets. This is ideal for getting synthesized answers rather than just search results. **Can I index custom documents into Glean programmatically?** Yes. Use `index_document` to add a single document with a JSON payload containing title, body, URL, and metadata. For large batches, use `bulk_index_documents` to index multiple documents in one request. Use `delete_document` to remove outdated content and `get_document` to verify indexing status.