oracle-ai-data-platform-workbench-engineer-agent — Claude Code plugin
A 37-skill agent for Oracle AI Data Platform Workbench: catalog discovery, Spark SQL, pipelines, governance and Delta Sharing, all driven by plain English.
/plugin install oracle-ai-data-platform-workbench-engineer-agent@claude-plugins-official
/plugin marketplace add anthropics/claude-plugins-officialWhat oracle-ai-data-platform-workbench-engineer-agent does
The plugin gives Claude an agent that covers the whole AIDP Workbench loop. It walks your catalog into a local grounding file at .aidp/catalog.md with join hints and a value dictionary per column. From there it writes Spark SQL you can run: CREATE, INSERT, UPDATE, DELETE, MERGE, OPTIMIZE/VACUUM, time-travel through DESCRIBE HISTORY. It also takes files in, profiles tables, sets quality rules, fixes cron pipelines, provisions Compute or AI Compute clusters and digs into jobs through the Spark UI. Governance is part of the same set: roles with per-resource permissions, masking, a credential store, Delta Sharing, audit logs, MLflow, Agent Flows across 13 node types with guardrails, and Knowledge Base RAG.
Control-plane calls go through the official aidp CLI when it is present; otherwise the agent sends oci raw-request to the same AIDP REST API. Interactive Spark SQL and notebook cells run through the bundled scripts/aidp_sql.py, because that needs the Jupyter WebSocket protocol and plain HTTP cannot do it.
Two things must exist before the first useful run. One is an oci CLI configured with a DEFAULT profile — either api_key or a token from oci session authenticate. The other is Python 3.x plus oci, requests, websocket-client and cryptography.
What it helps with
- turn the AIDP catalog into a grounding file with join hints and value lists
- convert an English question into Spark SQL that runs
- maintain Delta tables with compaction, cleanup and time-travel
- profile tables and apply data quality rules
- set per-resource permissions and masking
Example prompts for the agent
- Profile the store_sales table and suggest quality rules
- Share the gold schema with a recipient
How to install oracle-ai-data-platform-workbench-engineer-agent
Plugin marketplace
/plugin install oracle-ai-data-platform-workbench-engineer-agent@claude-plugins-official/plugin marketplace add anthropics/claude-plugins-officialStep by step, with commands for each platform: Claude Code plugins: marketplaces, installing, and building your own
Questions about oracle-ai-data-platform-workbench-engineer-agent
How do I install oracle-ai-data-platform-workbench-engineer-agent in Claude Code?
Add the marketplace first: /plugin marketplace add anthropics/claude-plugins-official. Then run /plugin install oracle-ai-data-platform-workbench-engineer-agent@claude-plugins-official.
What does oracle-ai-data-platform-workbench-engineer-agent do?
It runs Oracle AI Data Platform Workbench tasks in natural language: catalog exploration, Spark SQL, pipelines, clusters, governance and Agent Flows. Under the hood it uses the aidp CLI or oci raw-request.
Do I need to set up an MCP server?
No. What you need is the oci CLI with a DEFAULT profile and Python 3.x carrying four helper libraries for the bundled script.
