What it does
This MCP server connects an AI assistant to Tidebase, a system that gives AI agents their own identity and a credential vault. Agent API calls go through Tidebase, which injects the credential so the agent and model never see the key, and access can be scoped, audited and revoked. Tidebase also stores checkpoints, live state, queues, schedules and approval gates in your own Postgres, while your code keeps running in your own app or worker.
An assistant would reach for it to check the status of agent runs, decide on approval gates that are waiting for a human decision, or trigger recovery for a run that has stopped. Recovery relies on the stored checkpoints, so completed steps replay from storage rather than running again.
Running Tidebase requires a Postgres database. It can be started from a prebuilt server image, or from source with dependencies installed and the server and Studio running locally. Once the server is running, the MCP server is added to the assistant's configuration, and SDKs can connect to the same local server.
Limitations and requirements
Requires a Postgres database, which holds all Tidebase state. Running from source requires installing project dependencies. Tidebase does not run your code, so your application or worker process must still execute the work.
Safety
Our team has not reviewed this resource. Only automated rules checked it; resources with high-risk signals are not published. These are signals, not a guarantee that installing it is safe. Try it in a sandbox project first.
- Runs code or changes an assistant's behaviour
Sources
- Official MCP Registry first seen 13 Nov 2026 · last seen 13 Nov 2026 · its page on the source
io.github.BlueprintLabIO/tidebase
Inspect Tidebase agent runs, resolve approval gates, and trigger recovery from your AI assistant.
v0.1.1As shown on registry.modelcontextprotocol.io when we last read it