oleander Plugin for Claude and Codex
Willy Lulciuc
The best data agent experience requires context from tooling like MCP servers, skills, and CLIs. Today, we're excited to announce the official oleander plugin for Claude and Codex. Each plugin bundles the MCP server and every oleander skill into a single install.
The oleander architecture aligns with how agents query data:
- Fully-managed Iceberg catalog: no cluster to provision, serverless compute, live in minutes. Agents get a durable lake that scales with the query and stays ready for the bursty, on-demand workloads they create.
- Multi-engine smart routing: DuckDB, Spark, Polars, DataFusion. Picked automatically per query, based on size and cost history. No guessing.
- Cost and lineage on every query: what ran, what it cost, what changed. Tracked automatically by our context graph, powered by OpenLineage.
What you can do with it
- Explore from a conversation: drop a parquet or CSV, or point at a table, and ask questions in plain English.
- Query any dataset size: oleander routes to DuckDB, Spark, Polars, or DataFusion based on size and cost history.
- Bring your own catalog: connect an existing Iceberg catalog. Nothing to migrate.
- Debug a failed run: pull cost, what changed, and what it touched downstream from oleander's context graph.
- Investigate before it spreads: when an alert fires, lineage and schema history surface automatically.
- Schedule reports: set daily or weekly reports as Cowork tasks.
How to use it
To set up the oleander connector:
- [01]
Add the plugin marketplace
Open Cowork → Customize → + → Add marketplace from GitHub.
Paste
https://github.com/OleanderHQ/claude-plugin(orOleanderHQ/claude-plugin), then install the oleander plugin.Add and install the oleander marketplace in Cowork - [02]
Connect
Open Cowork → Customize → Plugins → oleander → Connectors → Connect.
Sign in with your oleander account and grant the permissions Claude requests.
Add oleander connector in Cowork - [03]
Verify
Ask your agent for a quick connection check:
“Are you connected to oleander?”
The agent should call
oleander_whoamiand return your organization context.
It offers the same MCP experience with simpler install across Cowork, Claude Code, and Codex.
Using a different agent? See Cursor and OpenCode setup.
Connect your data
Once the agent is linked, connect or upload at least one data source:
- Upload a file: parquet or CSV, no setup required.
- Connect your warehouse: Snowflake, BigQuery, and Iceberg catalogs under Settings → Lake.
To try it quickly, download the sample Iris dataset and upload it at oleander.dev/app/upload.
Then ask in plain English. For example:
- Analytics: "What's the average petal length in
default.iris, broken down by species?" - Exploratory: "Explore the iris dataset. What looks unusual across species?"
Next steps
We're excited to see how this connection makes it easier to build with oleander, Claude, or Codex. See the Quickstart, Claude setup, and Codex setup for everything oleander brings to your agent.
If you have any questions, please reach out and we'll be happy to help.