Redash MCP Server

Community

Connect AI assistants to Redash for data queries and dashboard management.

How to install ↓ Compare GitHub source repository ↗

Category: Database #41 of 58
Stars: ★ 105 Very popular
Transport: stdio (local process) · Streamable HTTP
Runtime: Node.js 22.13+ · Docker
Credentials: API key / credential required
License: MIT
Last commit: 9 days ago
Tools: 27

49FMRS · D

Based on the README: this is a non-official Redash MCP server supporting stdio and stateless Streamable HTTP, with broad query, schema, dashboard, and visualization management. Check API-key permissions, data exposure, and gateway authentication before deployment.

Strongest · Documentation 14/20 Weakest · Security and permissions 7/20

Reliability

8/20

Security and permissions

7/20

Maintenance

7/20

Documentation

14/20

Setup experience

13/20

Why each score

Reliability 8/20: The README claims stdio and HTTP transports plus query, dashboard, and visualization management, and lists test commands, but the server manifest is not cached and no source files, CI workflows, or coverage evidence are provided. Initialization, tool listing, and implementation behavior therefore cannot be verified. Startup depends on Node.js 22.13+, Corepack, pnpm, a Redash URL, and an API key; controllable error behavior is documented but not source-verifiable, so substantial points are deducted.
Security and permissions 7/20: The API key is configured through environment variables, the server manages the Authorization header and prevents override through extra headers, and HTTP defaults to loopback with Host/Origin allowlists. The README also explicitly states that these allowlists do not authenticate MCP clients and warns against direct internet exposure. However, REDASH_EXTRA_HEADERS are sent to every Redash request, HTTP mode has no built-in client authentication, and side-effecting operations such as archive_query and delete_visualization have no documented confirmation or fine-grained permission evidence. Source files are unavailable for verification, so the score is substantially reduced. No explicit red-line evidence from the prompt was found.
Maintenance 7/20: The README lists v1.0.0 and v1.1.0 and describes tests and Docker images; the repository is not archived and has seven open issues, but stars and issue counts are not maintenance proof. There is no evidence of commit cadence, release dates, issue response, dependency updates, security-response channel, or verifiable CI. The metadata says the license is unknown while the README claims MIT, creating an evidence inconsistency, so only a low-mid score is justified.
Documentation 14/20: The documentation covers installation, environment variables, stdio and Streamable HTTP, client configuration, Docker, telemetry, tool categories, parameters, pagination limits, BigQuery-specific behavior, health checks, and several operational edge cases. Points are deducted because the manifest and source are unavailable, so tool parameters and behavior cannot be checked against implementation. It also lacks a clear cost/permission model, systematic troubleshooting guidance, and a security-response process; license metadata conflicts with the README.
Setup experience 13/20: The README provides Node.js 22.13+, Corepack, pnpm, build, startup, and Claude Desktop configuration steps, plus Docker and HTTP examples, making the mainstream setup path understandable. Points are deducted because users must manually install the runtime, enable Corepack, configure a Redash instance, and supply an API key. There is no verifiable evidence of a published package, pinned release workflow, cross-platform validation, or CI; HTTP deployment additionally requires reverse-proxy authentication and allowlist configuration.

Static review · not runListed 2026-08-14

Read the FMRS scoring method →

Fit and risk

What it can accessUses the networkChanges third-party account data

Best for

Teams that already operate Redash and want MCP access to its analytics features.
MCP clients requiring query execution, schema discovery, and dashboard management.
Operators who need OpenTelemetry visibility into MCP and Redash requests.

Not for

Users without a Redash instance or API key.
Use cases requiring direct database access without the Redash API.
Deployments that would expose the HTTP listener directly to the internet.

Required permissions

Access to a Redash instance is required.
REDASH_API_KEY is required; the server manages the Authorization header using the Key scheme.
Depending on the tools used, the API key may permit reading, creating, updating, or archiving queries and managing dashboards and visualizations.

Risks and side effects

Query results, SQL, descriptions, visualizations, and CSV data may contain sensitive business information.
When OTel message-content capture is enabled, query arguments, results, and error content may be sent to telemetry backends.
HTTP Host and Origin allowlists do not authenticate MCP clients; run the server behind an authenticated gateway or reverse proxy.
Archiving is a soft delete, while visualization deletion and query, layout, or parameter updates still modify Redash state.

Setup

Before you start

Runtime：Node.js 22.13+ · Docker

REDASH_URL required Your Redash instance URL, e.g. https://redash.example.com.

REDASH_API_KEY requiredsecret Your Redash API key from Redash user settings; sent as the Authorization: Key header.

Other optional settings (19)

REDASH_TIMEOUT optional API request timeout in milliseconds, default 30000.

REDASH_MAX_RESULTS optional Maximum number of results to return, default 1000.

REDASH_EXTRA_HEADERS optional Extra HTTP headers added to every Redash request; JSON object string or semicolon/comma-separated key=value pairs.

REDASH_SOCKS_PROXY optional SOCKS proxy URL for routing requests, e.g. socks5h://localhost:1080.

MCP_TRANSPORT optional MCP transport: stdio, http, or streamable-http; default stdio.

MCP_HTTP_HOST optional Host for Streamable HTTP mode, default 127.0.0.1.

MCP_HTTP_PORT optional Port for Streamable HTTP mode, default 3000.

MCP_HTTP_PATH optional Streamable HTTP endpoint path, default /mcp.

MCP_HTTP_ALLOWED_HOSTS optional Comma-separated Host header allowlist for HTTP mode.

MCP_HTTP_ALLOWED_ORIGINS optional Comma-separated browser Origin hostname allowlist; also controls CORS.

OTEL_EXPORTER_OTLP_ENDPOINT optional OTLP exporter endpoint for traces/metrics/logs.

OTEL_METRICS_EXPORTER optional Metrics exporter, e.g. otlp, prometheus, or none.

OTEL_EXPORTER_OTLP_PROTOCOL optional OTLP protocol: grpc, http/protobuf, or http/.

OTEL_TRACES_EXPORTER optional Explicit traces exporter; none disables.

OTEL_LOGS_EXPORTER optional Explicit logs exporter; none disables.

OTEL_EXPORTER_PROMETHEUS_HOST optional Host for the standalone Prometheus listener.

OTEL_EXPORTER_PROMETHEUS_PORT optional Port for the standalone Prometheus listener.

OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT optional Set true to record tool arguments/results (including SQL and rows) in telemetry; omitted by default.

OTEL_SDK_DISABLED optional Set true to disable all telemetry signals.

Prepare Node.js 22.13 or later, Corepack, access to a Redash instance, and a Redash API key. Configure Claude Desktop with npx and set REDASH_URL and REDASH_API_KEY. Alternatively, clone the repository and run corepack enable, pnpm install, pnpm run build, and pnpm start. The default transport is stdio; set MCP_TRANSPORT=http for stateless Streamable HTTP.

claude_desktop_config.json

{"mcpServers":{"redash":{"command":"npx","args":["-y","@suthio/redash-mcp"],"env":{"REDASH_API_KEY":"your-api-key","REDASH_URL":"https://your-redash-instance.com"}}}}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json

{
  "servers": {
    "redash": {
      "command": "npx",
      "args": [
        "-y",
        "@suthio/redash-mcp"
      ],
      "env": {
        "REDASH_API_KEY": "your-api-key",
        "REDASH_URL": "https://your-redash-instance.com"
      }
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal

claude mcp add redash -e REDASH_API_KEY=your-api-key -e REDASH_URL=https://your-redash-instance.com -- npx -y @suthio/redash-mcp

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

After starting, check that tools like list_queries and execute_query appear in the client's tool list, then call list_queries to confirm it returns queries from your Redash instance.

Troubleshooting

Verify that REDASH_URL and REDASH_API_KEY are set and that the key can access the target Redash instance.
Verify Node.js is at least 22.13 and Corepack is enabled.
In stdio mode, ensure logs go to stderr and are not treated as MCP output.
In HTTP mode, check MCP_TRANSPORT, host, port, path, and Host/Origin allowlists; the default endpoint is 127.0.0.1:3000/mcp.
HTTP mode does not provide a standalone GET SSE stream; use POST /mcp.
For large schemas, paginate get_schema with page, pageSize, search, and nextPage.

Things to try

Once connected, you can ask your AI assistant things like:

List all queries in my Redash instance
Execute query 42 and get the results as CSV
Show me the schema of data source 7, filtering tables named analytics_public., 10 per page
Move widget 3 on the dashboard to row 0, column 4

Tools 27

list_queries read-only

List available queries in Redash.

get_query read-only

Get details for a specific query.

create_query writes

Create a query in Redash.

update_query writes

Update an existing query.

get_query_parameters read-only

Inspect saved query parameter definitions.

update_query_parameters writes

Update saved query parameter definitions.

archive_query destructive

Archive, or soft-delete, a query.

list_data_sources read-only

List available data sources.

Show 19 more tools

execute_query read-only

Execute a query and return results, with optional maxAge.

execute_parameterized_query read-only

Execute a saved parameterized query with typed value coercion, saved defaults, and optional maxAge.

execute_adhoc_query read-only

Execute an ad-hoc query without saving it to Redash.

get_query_results_csv read-only

Get query results as CSV, with optional refresh for current data.

get_schema read-only

Retrieve a data source schema with paginated tables and columns.

list_dashboards read-only

List available dashboards.

get_dashboard read-only

Get dashboard details and visualizations.

get_dashboard_layout read-only

Inspect dashboard widget positions, sizes, and visibility.

get_visualization read-only

Get details for a specific visualization.

get_dashboard_parameters read-only

Inspect dashboard parameter values and widget mappings.

update_dashboard_parameters writes

Update dashboard parameter values and ordering.

update_dashboard_layout writes

Move or resize multiple widgets in one call.

update_widget_layout writes

Move or resize a single widget.

get_widget_parameter_mappings read-only

Inspect a widget's parameter mappings.

update_widget_parameter_mappings writes

Update a widget's parameter mappings.

create_visualization writes

Create a visualization for a query.

update_visualization writes

Update an existing visualization.

update_chart_visualization writes

Update chart-specific options such as series type, column mapping, series options, legend, and axis settings.

delete_visualization destructive

Delete a visualization.

Use cases

Discovering and executing Redash queries through an AI assistant.

Running saved parameterized queries with typed values and saved defaults.

Exploring data-source schemas to support query writing.

Creating and maintaining queries, dashboard layouts, and visualizations.

Supported clients

Claude Desktop

Listed from the project's documentation, not tested by this site.

Overview

Redash MCP Server exposes Redash queries, data sources, schemas, dashboards, and visualizations through the Redash API to MCP clients. It supports ad-hoc and saved parameterized query execution, query creation, updating and archiving, schema discovery, dashboard layout and parameter mapping management, and chart-specific visualization updates.

Similar servers

MCP ClickHouse 79 · B

Connect ClickHouse to your AI assistants

★ 884 · Tools 4 Compare with this →

ChEMBL MCP Server 75 · B

Unofficial MCP server providing advanced access to the ChEMBL chemical database

★ 89 · Tools 27 Compare with this →

MCP Toolbox for Databases 74 · B

Google's official database MCP toolbox — define AI-accessible database tools via config

★ 16.6k Compare with this →

DBHub 74 · B

Bytebase's official, minimalist two-tool database MCP server

★ 3.6k · Tools 2 Compare with this →

Compare with other MCP servers

Source revision 2d8e2e544c37 Data synced 2026-10-12 Read the FMRS scoring method