InfluxDB 3 MCP Server
Official InfluxData MCP server: the agent reads InfluxDB 3 schemas and runs SQL and InfluxQL queries with output limits
- 38 ★ on the repository
- Last commit
Broad access
Why this level
- An admin token opens database deletion and token issuing, and regenerating the operator token is irreversible
- The agent sends SQL and InfluxQL with the token's permissions, and the token itself sets the access boundary
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add mcp/influxdb-3-mcp-serverDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- An admin token opens database deletion and token issuing, and regenerating the operator token is irreversible
- The agent sends SQL and InfluxQL with the token's permissions, and the token itself sets the access boundary
This is third-party code. Review the repository files before installing.
What it does
InfluxData's MCP server connects an agent to InfluxDB 3: Core, Enterprise, Cloud Dedicated, Clustered and Cloud Serverless. The agent lists databases and tables, inspects schemas, runs SQL and InfluxQL queries and checks instance health. Reads are built with model mistakes in mind. The query_sql and query_influxql tools cap the output, and each response carries a completeness flag, a row count and warnings. investigate_database samples each table over a time window, 24 hours by default, and keeps going if one table is unreachable. Errors arrive with a code, a hint for the next step and a flag saying whether a retry makes sense. There are also tools for writing and administration: write_line_protocol, creating and deleting databases, issuing and deleting tokens. The INFLUX_MCP_TOOL_PROFILE=readonly profile removes all of them, leaving the agent with exploration and queries only.
Who it is for. Analysts and engineers who keep metrics, telemetry or sensor data in InfluxDB 3 and want to ask questions about it in plain words from Claude Desktop, Cursor or Codex.
Good fit when
- A dashboard stopped showing data after a release and you need to find out whether the schema was renamed or the InfluxQL query broke
- You face an unfamiliar database and the agent has to map its tables and columns from data samples
- You want an answer like "which sensors had the highest average temperature in the last day" without writing SQL by hand
- You are preparing a batch of data: the agent writes line protocol rows into a test database
- You want to check connectivity and instance health before running scripts
Not a fit when
- You need cardinality estimates: the server does not compute them yet
- The agent must not hold a token that can write: use the readonly profile together with a read-only token
- Your database runs InfluxDB 1.x or 2.x: the server targets version 3
- You need continuous monitoring with alerts: the server answers agent requests and tracks nothing on its own
Example request
Which databases does this instance have, and which sensors had the highest average temperature over the last 24 hours?MCP
- Transport
- stdio
- Authentication
- API key
Environment variables
INFLUX_DB_INSTANCE_URLrequired- InfluxDB 3 instance URL
INFLUX_DB_TOKENrequired, secret- Admin token, scoped token or JWT
INFLUX_DB_PRODUCT_TYPErequired- Product type: core, enterprise, cloud-serverless, cloud-dedicated or clustered
INFLUX_MCP_TOOL_PROFILE- The value readonly keeps read-only tools
Limitations
You need a running InfluxDB 3 instance with its address and a token. Cloud Dedicated and Clustered add a cluster ID and management tokens. Running through npx needs a recent Node.js: the README lists minimum versions for the 22, 24 and 26 lines and warns that Node.js 20 is end of life.
Some tools are irreversible: deleting a database, deleting a token, regenerating the operator token. Tool limits steer the model, while the real access boundary is the permissions of the token itself. Version 2.0 changed the query response and health check formats, so clients that parse old responses need an update.
How to disable. Remove the influxdb block from the MCP client configuration and restart the client. Revoke any tokens the agent issued in InfluxDB itself.
FAQ
How do I connect InfluxDB 3 to Claude Desktop or Cursor?
Add a block to the client config with the npx command and the arguments -y @influxdata/influxdb3-mcp-server, and set INFLUX_DB_INSTANCE_URL, INFLUX_DB_TOKEN and INFLUX_DB_PRODUCT_TYPE (core, enterprise, cloud-serverless, cloud-dedicated or clustered) in env. Restart the client and call health_check.
How do I stop the agent from writing to the database?
Set INFLUX_MCP_TOOL_PROFILE=readonly: the profile does not load tools for writing, database administration, token management or host execution. Token permissions remain the main protection, so use a read-only token.
Which InfluxDB versions does the server support?
InfluxDB 3 Core and Enterprise, plus Cloud Dedicated, Clustered and Cloud Serverless. INFLUX_DB_PRODUCT_TYPE selects the product, and some tools exist only on certain products: resource tokens, for example, only on Enterprise.
What does the agent see when a query fails?
A structured error with the fields code, recovery and retryable. A 403 arrives as not_found_or_denied whether or not the database exists, and a query naming an unknown table gets a hint to call list_tables or describe_table.
Does the server run anything on my machine?
The server itself runs on your computer over stdio and talks only to your InfluxDB instance address. The readonly profile also removes the tools that execute on the host.
Discussion
No questions or reviews yet.
Related
MCP serversAn agent queries GreptimeDB metrics, logs and traces with SQL, TQL and RANGE, with write protection and data masking
MCP server that gives an agent access to MySQL: schema inspection and queries, read-only by default, writes behind flags
InfraNodus MCP server turns any text into a knowledge graph and shows topic clusters, the gaps between them and research questions
One MCP server for SQLite, SQL Server, PostgreSQL and MySQL: read and change data, edit schemas and export to CSV and JSON
MySQL MCP Server Pro
mysql_mcp_server_pro
A MySQL MCP server that runs SQL under access roles, finds tables by comments and analyzes locks, indexes and query plans
Official Apache IoTDB MCP server: SQL over time series in tree and table models, with CSV and Excel export
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| InfluxDB 3 MCP Server (this card) | yes | API key | 38 | |
| Prometheus MCP Server | no | not required | 525 | PromQL queries over Prometheus metrics |
| ClickHouse MCP Server | no | API key | 904 | the ClickHouse columnar analytics database |
| Grafana MCP Server | yes | API key | 3,663 | Grafana dashboards and data sources |
README in short
The README covers prerequisites, read-only workflows, scenario examples and a table of 27 tools marking which products support each. It also lists four resources and three prompts, plus setup for every instance type. The server runs through npx, from source and in Docker.