# Typesense Cloud MCP > Typesense Cloud MCP lets your AI agent manage and debug your fast search infrastructure directly through chat. Check cluster health, track real-time performance metrics, list API keys, or run complex multi-searches across multiple collections without ever touching a terminal command line. It gives you full operational visibility into every part of your distributed search stack. ## Overview - **Category:** loved-by-devs - **Price:** Free - **Tags:** search-engine, cluster-monitoring, latency-tracking, api-key-management, federated-search ## Description This MCP connects your Typesense Cloud endpoint to any AI agent, giving you hands-on control over your lightning-fast search infrastructure right from your chat window. Instead of digging through confusing CURL outputs or complex dashboard views to diagnose slow searches, you simply ask your agent what's wrong. You can verify if all nodes are online and ingesting data smoothly, measure latency spikes against real usage logs, or even execute multi-search commands across several collections simultaneously. When you connect this MCP via Vinkius, you get a single pane of glass for cluster forensics. This eliminates the need for deep knowledge of terminal diagnostics; you just delegate those complex operational checks directly to your highly capable AI. ## Tools ### get_cluster_health Checks the operational health status of your entire Typesense search cluster. ### get_cluster_metrics Retrieves detailed performance and usage metrics for the Typesense cluster, including resource consumption. ### list_collection_aliases Lists all virtual aliases that map to your real search data collections. ### list_collections Lists every active search collection housed within the cloud cluster. ### list_api_keys Shows a full list of all API keys configured and used by the Typesense cluster. ### execute_multi_search Runs several different, complex search requests in one single API call using JSON arrays. ## Prompt Examples **Prompt:** ``` Check the cluster health to verify Typesense is up in London. ``` **Response:** ``` The cluster responded successfully in 12ms. Status is 'ok'. Your Typesense nodes are fully operational. ``` **Prompt:** ``` List all active collections inside this database environment. ``` **Response:** ``` You have 3 core collections currently structured: 1. 'global_docs' (150,000 docs) 2. 'commerce_catalog' (2,000 docs) and 3. 'test_index_v1' (0 docs). Need the exact schema for the e-commerce one? ``` **Prompt:** ``` Fetch the performance metrics of the cluster and tell me if response times are above 100ms. ``` **Response:** ``` Pulling cluster metrics... The search_latency median is currently at 14ms globally. The 99th percentile spike reached 88ms. Fortunately, there are no endpoints dropping past 100ms. CPU stands at an idle 12% usage. ``` ## Capabilities ### Check overall cluster status Determines if all nodes are running and accessible, confirming uninterrupted data ingestion. ### Retrieve performance measurements Gathers real-time metrics like usage logs, active search workloads, and resource consumption patterns. ### Run complex searches across multiple sources Sends a single request to perform simultaneous multi-searches against several defined collections. ### Manage virtual collection names Lists all aliases that abstract the concrete structure of your search data. ### Audit configured access keys Retrieves a list of every API key currently set up for the cluster. ## Use Cases ### The search results are suddenly slow after deployment A developer asks their agent to check the performance metrics, running `get_cluster_metrics`. The agent reports that the 99th percentile spike hit 88ms, immediately pointing out a specific bottleneck in the resource usage before any code needs to be rolled back. ### I need to validate three different data sets for an A/B test A search engineer uses `execute_multi_search` to hit three distinct collections (e-commerce, global docs, and archived content) in one API call. This confirms the query structure works across all required sources before any UI development begins. ### We think a key might have been compromised A sysadmin asks the agent to run `list_api_keys`. The tool returns an exhaustive list, allowing them to immediately audit and revoke specific keys without manually navigating the cloud console. ### We need a full inventory of our search data sources A DBA runs the agent command for `list_collections` and receives a clean list of all active collections. They can then use `list_collection_aliases` to understand how those virtual names are exposed publicly. ## Benefits - Quickly diagnose failures: Instead of manually checking status codes, use the `get_cluster_health` tool to instantly confirm if your nodes are online and fully operational. - Pinpoint slow spots: The `get_cluster_metrics` tool provides usage logs and latency thresholds, letting you see exactly when and why performance degrades. - Complex querying made easy: Don't write multi-search logic; just run `execute_multi_search` to query multiple collections simultaneously in one go. - Maintain clean architecture: Use `list_collection_aliases` to track which virtual names are pointing where, keeping your data structure auditable and clear. - Secure access management: The `list_api_keys` tool gives you a single source of truth for auditing every key used against the cluster. ## How It Works The bottom line is that you get immediate answers about your search cluster status without needing to write or execute complex command-line code. 1. Subscribe to this MCP and provide your active Typesense Host URL along with your specific API Key. 2. Connect your agent client (like Cursor or Claude) to the Vinkius catalog using these credentials. 3. Ask your AI client a question, like 'What's the current search latency?' The agent executes the necessary tool call and gives you the direct answer. ## Frequently Asked Questions **How do I check the overall health of my Typesense cluster using the Typesense Cloud MCP?** Run the `get_cluster_health` tool. This immediately verifies if all nodes are online and reachable, telling you right away if there's a systemic outage or just a minor hiccup. **Can I use Typesense Cloud MCP to test multiple collections at once?** Yes, the `execute_multi_search` tool allows you to send one single request that targets and searches across several different data collections simultaneously. This is perfect for A/B testing. **What should I run if my search performance seems inconsistent?** Start by using `get_cluster_metrics`. This tool pulls usage logs, active workloads, and latency thresholds to help you pinpoint if the issue is CPU load or network throttling. **How do I find out what my virtual data names are?** Use `list_collection_aliases` to see all the abstract aliases. This tells you which virtual name clients should use, even if the underlying collection structure changes. **Is there a way to audit who has access via Typesense Cloud MCP?** Run `list_api_keys`. This tool provides an inventory of every API key currently configured for your cluster, which is essential for security audits and managing permissions.