Skip to content

About

This repo act an aggregator and a single entry points for all the other repos containing blueprints

Resources

Stars

1 star

Watchers

0 watching

Forks

Latest commit

 

History

7 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 

Repository files navigation

OCI Agents Blueprints

Practical blueprints for bringing enterprise AI agents to Oracle Cloud Infrastructure.
From an idea to a deployable service: knowledge retrieval, interoperability, tools, and durable workflows.

Oracle Cloud Infrastructure Enterprise AI Spec-driven development Cloud-ready deployment

One entry point, six core patterns. Choose the blueprint that fits your use case and start from a readable, testable implementation designed for OCI.

Why this collection

Building an agent for a demo is relatively simple. Making it useful in an enterprise context requires clear integration boundaries, secure configuration, observability, tests, and a repeatable deployment path.

This GitHub collection brings together independent, complementary blueprints for those challenges. Each repository provides reference code, operational documentation, and concrete extension points. This is not a monolithic platform: it is a toolkit of patterns that can be combined to fit the architecture you need.

                         ┌─────────────────────────┐
                         │  OCI Agents Blueprints   │
                         │       entry point        │
                         └────────────┬────────────┘
      ┌──────┬──────┬──────┬──────┬──────┬──────┐
      ▼      ▼      ▼      ▼      ▼      ▼      ▼
 📚 Knowledge 🔄 Agents 🧰 Tools 🧠 Memory 💾 Durable 📈 Observe ✨ New
     RAG       A2A      MCP             state       blueprints

Available blueprints

Blueprint When to use it What you will find
📚 OCI RAG Agent Blueprint You need an assistant that answers from enterprise documents, policies, manuals, or a knowledge base. OCI Vector Store and Responses API integration, a FastAPI backend, conversations and streaming, CLI, Next.js UI, Docker Compose, and an Agent Factory for deployment.
🔄 OCI LangGraph A2A Blueprint You want to expose a LangGraph workflow as an interoperable service that other agents and orchestrators can invoke. A LangGraph agent with typed state, an A2A 1.0 HTTP/SSE server, Agent Card, test clients, containerization, and OCI deployment with optional JWT protection.
🧰 OCI MCP Blueprint You want to create and publish tools, resources, and prompts through the Model Context Protocol. A Python FastMCP skeleton, sample tools, a Streamable HTTP client, centralized configuration, tests, Docker Compose, and a deployment factory for OCI Enterprise AI Hosted Applications.
🧠 OCI Agent Memory Blueprint You need an agent to retain relevant context, preferences, and facts across interactions with clear scope and governance. Oracle Agent Memory backed by Autonomous Database, progressive examples for thread and message memory, configuration guidance, and patterns for short- and long-term memory.
🛡️ OCI Guardrails Blueprint You need to apply consistent safety controls to an agent's inputs and model responses. OCI Guardrails patterns with LangChain and langchain-oci, including configuration, testing, policy boundaries, and production-oriented safety practices.
💾 OCI LangGraph Checkpoint Blueprint You have long-running or human-in-the-loop agent processes that must survive pauses, restarts, and recovery. LangGraph checkpoints persisted in Oracle Autonomous Database, progressive examples, FastAPI/SSE APIs, human approvals, thread resumption, and operational guidance.
📈 OCI Observability Blueprint You need to understand, troubleshoot, and operate AI agents running on OCI. Langfuse and OCI APM instrumentation patterns for agent and LLM traces, request correlation, token and latency telemetry, error analysis, safe data capture, and operational best practices.
✨ Future blueprints You need an OCI agent pattern that is not yet covered. The collection is extensible: new repositories for complementary use cases and agentic patterns will be added here.

What they have in common

Although they solve different problems, the blueprints share the same principles:

  • OCI at the core: integrations and deployment paths designed for Oracle Cloud Infrastructure and Enterprise AI.
  • From local to cloud: local run and validation instructions, container images, and repeatable flows to OCI Container Registry and Hosted Applications where applicable.
  • Explicit contracts: APIs, protocols, and configuration are documented to keep integrations predictable.
  • Spec-driven development: meaningful features begin with specifications, acceptance criteria, and tests.
  • Ready to adapt: concrete, modular examples that you can use as a foundation by replacing the domain, tools, models, and interfaces with your own.
  • Responsible configuration: secrets and environment-specific values stay out of source code and are managed with environment variables and ignored local files.

How to choose

Start with your most important architecture question:

  • Do I need reliable answers from documents? → choose the RAG blueprint.
  • Do I need agents to collaborate or integrate with a standard orchestrator? → choose LangGraph A2A.
  • Do I need to make tools and context available to an agent or compatible client? → choose MCP.
  • Do I need an agent to retain relevant user or application context over time? → choose Agent Memory.
  • Do I need to pause, resume, and recover an agentic process? → choose LangGraph Checkpoint.
  • Do I need to trace, debug, measure, and operate an agent in production? → choose Observability.

The patterns can coexist. An A2A agent can use MCP tools, a LangGraph workflow can maintain durable checkpoints, an agent can use Oracle Agent Memory for relevant long-term context, and the Observability blueprint can trace and monitor any of these components. A RAG assistant can also become a capability in a wider agent system.

Start here

Open the repository that is closest to your use case, follow its quickstart, and customize its documented extension points. Each blueprint maintains its own README, specifications, and operational guides, keeping this entry point simple while the implementation details remain close to the code.


Build useful agents today. Design them to evolve tomorrow.

About

This repo act an aggregator and a single entry points for all the other repos containing blueprints

Resources

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors