MCP observability, served warm.

See what your MCP
server actually does.

Tools get called. Clients come back. Something breaks.
Dashi shows you the whole story.

Open source today. Hosted workspaces coming later.

Meet Dashi

From “it works” to knowing how.

See adoption, tool calls, latency and context costs together. Follow real activity across your MCP server, with the data in your own Postgres.

Dashi dashboard shown with demonstration data.

A few lines. A much clearer picture.

Add an SDK to the MCP server you already run.

Set DASHI_URL and DASHI_KEY from your collector project. Events ship in the background.

Open Dashi. Follow a call from connection to result.

pip install "dashi-mcp[mcp]"

from dashi_mcp import Dashi

dashi = Dashi.from_env()     # reads DASHI_URL and DASHI_KEY
dashi.instrument(server)

Quick start from the Python SDK README.

Coming soon

A live MCP check for Claude and ChatGPT. Clear score cards, actionable rules.

Your infrastructure. Or ours, soon.

Apache-2.0. Your Postgres, your events, your infrastructure. Run the collector and dashboard today.

Create your account now. Hosted Dashi workspaces arrive in a later phase; these are the live free-plan limits.

Loading plan limits…