Creativity · MCP — server

MCP TimescaleDB Server

TimescaleDB extends PostgreSQL with hypertables, continuous aggregates, and compression for time-series workloads. The MCP TimescaleDB Server surfaces that functionality as tools — list hypertables, describe chunks, run analytical SQL — to any MCP client, making it easy for agents to investigate metrics and event data.

MCP facts

Kind: server
Ecosystem: anthropic-mcp
Language: Python
Transports: stdio

Capabilities

Tools: list_hypertables, describe_hypertable, list_continuous_aggregates, run_query
Works against Timescale Cloud, self-managed, or embedded setups
Honors a read-only connection role

Install

pip install mcp-timescale

Configuration

{
  "mcpServers": {
    "timescale": {
      "command": "uvx",
      "args": ["mcp-timescale"],
      "env": {
        "TIMESCALE_URL": "postgres://tsdbadmin:<password>@<host>:<port>/tsdb"
      }
    }
  }
}

Frequently asked questions

How is this different from the generic Postgres MCP server?

It exposes Timescale-specific introspection: hypertables, chunks, continuous aggregates, retention policies — plus smart defaults for time-bucketed queries.

Does it handle continuous aggregates?

Yes — it lists continuous aggregate views so the model can prefer them over raw hypertables for large time ranges.

Best practices?

Use a read-only role, cap statement_timeout, and prefer querying continuous aggregates or compressed chunks to keep token usage low.

Sources

TimescaleDB Documentation — accessed 2026-04-20
Model Context Protocol — accessed 2026-04-20