ceven

ClickHouse

Runs complex SQL queries across your column oriented data stores to extract real time insights and mirrors analytics results into your operational tools.

Try ClickHouse in Ceven

Ask Ceven anything
Standard

Why use Ceven?

  1. AI native ClickHouse integration

    • Describe the outcome and Ceven picks the right ClickHouse calls, fills the parameters, and checks the result.
    • Structured, agent friendly tool schemas so each call runs reliably instead of by guesswork.
    • Rich coverage for reading, writing, and querying your ClickHouse data, across all 6 of its actions.

  2. Managed auth

    • Built in OAuth with automatic token refresh and rotation.
    • One place to manage, scope, and revoke ClickHouse access.
    • Per user and per environment credentials instead of shared keys.

  3. Agent optimized design

    • Actions are tuned from real success and error rates so reliability climbs over time.
    • Full execution logs so you always know what ran in ClickHouse, when, and on whose behalf.
    • The agent pauses and asks when ClickHouse is unclear instead of plowing ahead.

  4. Enterprise grade security

    • Fine grained access so you control which agents and people can reach ClickHouse.
    • Least privilege by default, read scopes first and only the writes a workflow needs.
    • A full audit trail of every ClickHouse action to support review and sign off.

Supported tools

Every action Ceven's agents can run on ClickHouse, and when to use it.

Execute query
Use this to run a SQL query against your ClickHouse database to pull specific metrics, aggregates, or raw event rows.
Get database schema
Pull a complete overview of the database including all tables and column definitions to understand the data model.
Get table schema
Pull detailed column types, keys, and sample data for a specific table before constructing a complex join.
List databases
Pull a list of all available databases in the ClickHouse instance to identify the correct data source.
List tables
Pull all tables within a database including their engine type, storage size, and total row count.
Run aggregate query
Use this for high speed summations or averages across billions of rows using ClickHouse aggregate functions.
Insert event data
Push a batch of event records into a table. Use this when mirroring data from another SaaS tool into ClickHouse.
Update table schema
Add a new column or modify an existing column type to accommodate new event properties.
Create table
Initialize a new table with a specific engine like MergeTree for optimized analytical performance.
Drop table
Remove an obsolete table or a temporary staging table after a data migration is complete.
Check query profile
Pull the execution profile of a slow query to identify bottlenecks in the column scans.
Search table names
Query the system tables to find all tables matching a specific naming pattern or tag.
Execute ClickHouse Query
Execute a sql query in clickhouse and return the results. this is the primary action for querying data from clickhouse databases.
List ClickHouse Databases
List all databases in the clickhouse instance. useful for discovering available databases before querying tables.
List ClickHouse Tables
List tables in clickhouse databases. returns information about tables including their engine, size, and row count.

15 actions · scroll to see them all

Frequently asked questions

Ceven implements a streaming cursor approach for large ClickHouse result sets to prevent memory overflow in the workflow layer. When a query returns more than ten thousand rows, the agent automatically paginates the results and processes them in chunks. You can specify a limit in your prompt to keep responses concise, or allow the agent to stream the full set into a destination like a Google Sheet or an S3 bucket. This ensures that even if you query a table with billions of rows, the agent only holds the necessary window of data in active memory while it performs the requested transformation or delivery.
Yes, Ceven can perform schema migrations if you grant it the appropriate write and manage permissions. The agent can add columns or create new tables using the MergeTree engine to optimize for your specific query patterns. We recommend using a restricted user for these actions in production environments. The agent always pulls the current schema before attempting an alter command to ensure the change is compatible with existing data. You can also set up a human approval step in the Ceven workflow so a database administrator must sign off on any schema changes before they are executed.
Ceven optimizes queries for column oriented storage by encouraging the use of filtered scans and avoiding select star statements. However, heavy analytical queries can still consume significant CPU on your cluster. To mitigate this, you can configure ClickHouse user profiles to limit the maximum memory usage and execution time for the Ceven user. The agent also caches schema definitions to reduce the number of metadata calls. If you notice latency, we suggest creating materialized views for your most frequent reports, which Ceven can then query directly for near instant response times without scanning raw logs.
Ceven supports both ClickHouse Cloud and self hosted installations. For Cloud users, we connect via the HTTPS interface using secure credentials. For self hosted clusters, you can connect via the same HTTP interface or through a secure proxy. The agent automatically detects the version of ClickHouse you are running and adjusts the SQL syntax accordingly, as certain functions and table engines vary between versions. Whether you are using a managed service or your own Kubernetes deployment, the workflow experience remains the same, provided the network path is open for our agent to reach your endpoint.
One critical quirk is that ClickHouse does not support traditional transactional updates or deletes in the same way a row store like PostgreSQL does. Deletes and updates are handled as mutations, which are asynchronous and can be resource intensive if run frequently. If your workflow requires constant single row updates, you will see a performance drop or high CPU usage. We recommend using the ReplacingMergeTree engine for these use cases. Ceven is aware of this behavior and will suggest using a versioning column or a separate state table if it detects you are trying to perform frequent row level mutations.
Ceven uses industry standard encryption for all data in transit between our platform and your ClickHouse instance. We support TLS for all HTTP connections to ensure your queries and results are not intercepted. Credentials are encrypted at rest using AES 256 and are only decrypted in a secure execution environment at the moment the query is sent. We do not store the results of your queries permanently unless you explicitly map them to a long term storage destination. You can rotate your ClickHouse credentials at any time, and the agent will prompt you to update the connection once it detects a failure.
Yes, the agent can design and deploy materialized views to speed up your most common analytical paths. If you find yourself asking the same aggregate question every morning, you can tell Ceven to create a materialized view that pre calculates those totals. The agent will analyze the source table, suggest the optimal aggregation keys, and execute the create statement. This shifts the computational load from query time to ingestion time, allowing your dashboards and alerts to load instantly. The agent can also manage the lifecycle of these views, including refreshing them or updating the logic as your data needs evolve.
Absolutely. The best way to control access is through the native ClickHouse role based access control system. We recommend creating a dedicated Ceven user with read only access to specific databases or tables. The agent will only be able to see and query what that user is permitted to access. When the agent runs the list tables command, ClickHouse filters the output based on these permissions. This ensures that sensitive tables, such as those containing raw PII or internal system logs, remain completely invisible to the AI agent and cannot be accessed by any workflow author.

Alternatives to ClickHouse

Other tools that solve a similar problem. Ceven supports these too, so you can switch or run more than one at once.

Snowflake logoSnowflakeApache Druid logoApache DruidClickHouse Cloud logoClickHouse CloudDuckDB logoDuckDB

Try Ceven on your stack

Plug Ceven on top of the tools you already run. Connect ClickHouse and the rest of your stack, describe the outcome, and its agents handle the work end to end, days of it in minutes.

Get started for free