Commands come from the project's own docs or registry entry. Check them against the source before you run them.
Loading poll...
Metabase is an open-source business intelligence and embedded analytics tool, and this entry is the MCP endpoint its server exposes at your own host under /api/mcp, over streamable HTTP. The install line is claude mcp add --transport http metabase https://{metabase_host}/api/mcp, with the host in braces for you to fill. Because the endpoint lives on your Metabase, it is a self-hosted route and the agent sees only what that instance exposes. The licence in our data is a non-standard licence, so read Metabase's terms if you plan to embed it. The parent repository shows 50 thousand stars. We would suggest it to teams that already run Metabase and want people to ask questions of their data through an agent. The main limit is permissions: the agent acts with the access of the account it uses, so give it a read-only role. We have not tested it.
PostHog describes itself on GitHub as a platform for building products, with AI observability, analytics, session replay and feature flags, and this entry is its hosted MCP server at mcp.posthog.com over streamable HTTP and SSE.
OpenMetadata, per its repository, is an open platform for building trusted data context and business semantics for humans and AI, and this entry is the MCP endpoint at your own host under /mcp, over streamable HTTP.
DBX is a 25 MB cross-platform database client for more than 100 databases, including MySQL, PostgreSQL, SQLite, Redis, MongoDB, DuckDB and SQL Server, according to its GitHub description.
Tried it with one of these? Vote below: works for me, or not.
Questions
How do I install Metabase MCP?
Run this command: claude mcp add --transport http metabase https://{metabase_host}/api/mcp. The install box on this page has the command for each agent it works with.
Is Metabase MCP free?
It is published under the Other licence.
Who maintains Metabase MCP?
It is maintained by metabase. The code is at github.com/metabase/metabase.
Scrapling is a Python web scraping framework from D4Vinci, described on GitHub as handling everything from a single request to a full-scale crawl, and this entry is its MCP server.
Claude Flow is published on npm as claude-flow, and its repository now sits at ruvnet/ruflo, which the GitHub description calls the original agent harness for deploying multi-agent swarms and coordinating autonomous workflows.
Context7 is an MCP server from Upstash that serves up-to-date library documentation to coding agents, so the model reads current docs instead of guessing from training data.