skills.rest
oracle-devrel
oracle-devrelOFFICIAL
Explore all skills by this author →
WebsiteGitHub

build-paths-intermediate

Scaffold a Grok-4 tool-calling agent over an Oracle schema with MCP and ONNX embeddings.

4.3k|807|Updated Jan 16, 2024
Search Tags:#rag#mcp#oracle#onnx#open webui#in-db embeddings#sql agent
One-click install
npx skills add https://github.com/oracle-devrel/oracle-ai-developer-hub --skill build-paths-intermediate
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: build-paths-intermediate
Source: https://github.com/oracle-devrel/oracle-ai-developer-hub/tree/main/build-paths/intermediate
Command: npx skills add https://github.com/oracle-devrel/oracle-ai-developer-hub --skill build-paths-intermediate

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Build a production-shaped SQL tool-calling agent and in-database embedding workflow on Oracle, without the usual glue-code and integration pitfalls from RAG and chatbot prototypes.

Core Features & Use Cases

  • LLM-to-live-SQL via MCP: Lets Grok-4 choose tools like list_tables, describe_table, describe_schema, run_sql, and vector_search to answer questions grounded in your Oracle schema.
  • In-DB ONNX embeddings: Registers an ONNX MiniLM model in Oracle and uses VECTOR_EMBEDDING inside the database for both ingestion and query-time embedding.
  • Oracle-friendly full stack: Scaffolds a FastAPI adapter and Open WebUI so users can test the agent immediately with a durable chat-history layer.

Quick Start

Tell the AI to scaffold the intermediate path in your current empty directory, and ask it to build an NL2SQL demo with in-DB ONNX embeddings and Grok-4 tool calling over the Oracle schema.

Frequently Asked Questions about build-paths-intermediate

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build an NL2SQL agent with live Oracle database access using MCP?▼

Scaffold a Grok-4 tool-calling agent over an Oracle schema using oracle-database-mcp-server, which exposes tools like run_sql and describe_table for live SQL access and natural language queries.

How do I generate in-database embeddings with ONNX in Oracle 26ai?▼

Register a MiniLM ONNX model via onnx2oracle, then use the VECTOR_EMBEDDING function inside Oracle 26ai for both data ingestion and query-time embedding without external APIs.

Can I use Open WebUI to test an Oracle SQL tool-calling agent?▼

Yes, you can test the agent in Open WebUI by scaffolding a FastAPI adapter that wraps the MCP tools, providing a durable chat-history layer for immediate interactive testing.

Do I need an external embedding API for RAG workflows on Oracle Database?▼

No external embedding API is needed because the system uses in-database ONNX embeddings via OCI GenAI and the VECTOR_EMBEDDING function to handle both ingestion and query-time RAG workflows.

What are the limitations of using in-DB ONNX embeddings for SQL agent workflows?▼

This intermediate path requires existing RAG knowledge, manual ONNX model registration via onnx2oracle, and setting the OCI_GENAI_API_KEY, making it unsuitable for beginners seeking a no-code setup.

Related Skills

Explore More Software Engineering Skills→
diegoliriodiegolirio

sql-extract-ddl-oracle-mcp

Export Oracle schema DDL via MCP into Flyway-compatible SQL migrations.

Community
Advanced
oracle-devreloracle-devrel
4.3k

build-paths-beginner

Scaffold a RAG chatbot with OracleVS embeddings and a Grok-4 chat UI.

Official
Advanced
oracle-devreloracle-devrel
4.3k

build-paths

Route Oracle AI Database projects to matching build-path scaffolds by difficulty.

Official
Intermediate
ollieb89ollieb89

using-vector-databases

Implement vector database functionality for semantic search and RAG systems.

Community
Advanced