Wojciech-Rebisz · Feb 25, 2026 · Feb 24, 2026
@@ -1,110 +1,165 @@
+<div style="text-align: center;">
+
 # OpenAI Responses Agent
 
-Agent **without any agentic framework**: uses only the **OpenAI Python client** and **pure Python** (Responses API). No LlamaStack, LangChain, LlamaIndex, etc. – to show it can be done without frameworks. Uses `AIAgent` with chat, tools, and Action/Observation loop. Compatible with OpenAI API or any OpenAI-compatible endpoint (e.g. `BASE_URL` override). Python 3.12+.
+</div>
 
-# Use Agent Locally
+Agent **without any agentic framework**: uses only the **OpenAI Python client** and **pure Python** (Responses API). No LlamaStack, LangChain, LlamaIndex, etc. Uses `AIAgent` with chat, tools, and Action/Observation loop. Compatible with OpenAI API or any OpenAI-compatible endpoint (e.g. `BASE_URL` override). Python 3.12+.
 
-### Installation
+---
 
-```bash
-git clone <repository-url>
-cd Agentic-Starter-Kits
-```
-```bash
-python -m venv .venv
-source .venv/bin/activate  # On Windows: .venv\Scripts\activate
-```
+### Preconditions
 
-If you want to install Ollama: [Ollama site](https://ollama.com/) or [Brew](https://formulae.brew.sh/formula/ollama#default).
+- Copy/paste the `.env` file and set values for your environment
+- Choose **local** or **RH OpenShift Cluster** and fill the needed values
+- Run `./init.sh` to load values from `.env` into the environment
 
-**Install agent dependencies** (OpenAI client):
+Copy `.env` file:
 
 ```bash
-cd agents/base/openai_responses_agent
-uv pip install -e .
-# or: poetry install
+cp template.env agents/base/openai_responses_agent/.env
 ```
 
-### Setup Instructions
+#### Local
+
+Edit the `.env` file with your local configuration:
+
+```
+BASE_URL=http://localhost:8321
+MODEL_ID=ollama/llama3.2:3b
+API_KEY=not-needed
+CONTAINER_IMAGE=not-needed
+```
 
-**Option A – OpenAI API**
+> **Local setup (Ollama):** Port and model name can differ depending on your setup (e.g. Llama Stack on port 8321 vs Ollama on 11434, or a different model ID). Check your running services and `run_llama_server.yaml` (if using Llama Stack) and set `BASE_URL` and `MODEL_ID` accordingly.
 
-Create `.env` in the agent directory (or use repo `template.env`):
+Or for **OpenAI API** directly:
 
-```env
+```
 BASE_URL=https://api.openai.com/v1
 MODEL_ID=gpt-4o-mini
 API_KEY=sk-...
+CONTAINER_IMAGE=not-needed
 ```
 
-**Option B – Local OpenAI-compatible endpoint (e.g. Ollama)**
+#### OpenShift Cluster
 
-1. Pull model and run Ollama (or another OpenAI-compatible server).
-2. In `.env` set e.g. `BASE_URL=http://localhost:11434/v1`, `MODEL_ID=llama3.2`, `API_KEY=not-needed` (if not required).
+Edit the `.env` file and fill in all required values:
 
-**Configure Environment Variables**
+```
+API_KEY=your-api-key-here
+BASE_URL=https://your-llama-stack-distribution.com/v1
+MODEL_ID=llama-3.1-8b-instruct
+CONTAINER_IMAGE=quay.io/your-username/openai-responses-agent:latest
+```
+
+**Notes:**
+
+- `API_KEY` – contact your cluster administrator
+- `BASE_URL` – should end with `/v1`
+- `MODEL_ID` – contact your cluster administrator
+- `CONTAINER_IMAGE` – full image path where the agent container will be pushed and pulled from. The image is built locally, pushed to this registry, and then deployed to OpenShift.
+
+  Format: `<registry>/<namespace>/<image-name>:<tag>`
+
+  Examples:
 
-Copy the template (from repo root: `template.env`) or create `.env` in the agent directory:
+  - Quay.io: `quay.io/your-username/openai-responses-agent:latest`
+  - Docker Hub: `docker.io/your-username/openai-responses-agent:latest`
+  - GHCR: `ghcr.io/your-org/openai-responses-agent:latest`
+
+Go to agent dir:
 
 ```bash
-cp ../../../template.env .env
+cd agents/base/openai_responses_agent
 ```
 
-Edit `.env` with `BASE_URL`, `MODEL_ID`, and `API_KEY` as above.
+Make scripts executable:
+
+```bash
+chmod +x init.sh
+```
 
-**Run the example**
+Load values from `.env` into environment variables:
 
 ```bash
-cd examples
-python execute_ai_service_locally.py
+./init.sh
 ```
 
-**⚡ Or with [uv](https://docs.astral.sh/uv/)** (from repo root):
+---
+
+## Local usage (Ollama + LlamaStack Server)
+
+Create package with agent and install it in venv:
 
-1. Create venv and activate:
 ```bash
-uv venv --python 3.12
-source .venv/bin/activate
+uv pip install -e .
+uv pip install ollama
 ```
 
-2. Copy shared utils into the agent package:
+Install Ollama from the [Ollama site](https://ollama.com/) or via Brew:
+
 ```bash
-cp utils.py agents/base/openai_responses_agent/src/openai_responses_agent_base/
+# brew install ollama
+# or
+curl -fsSL https://ollama.com/install.sh | sh
 ```
 
-3. Install agent (editable) and its requirements:
+Pull required models:
+
 ```bash
-uv pip install -e agents/base/openai_responses_agent/. -r agents/base/openai_responses_agent/requirements.txt
+ollama pull llama3.2:3b
+ollama pull embeddinggemma:latest
 ```
 
-4. Run the example:
+Start Ollama service:
+
 ```bash
-uv run agents/base/openai_responses_agent/examples/execute_ai_service_locally.py
+ollama serve
 ```
 
-# Deployment on Red Hat OpenShift Cluster
+> **Keep this terminal open!**  
+> Ollama needs to keep running.
 
-### Step 1: Initialize the Agent
+Start LlamaStack server:
 
 ```bash
-cd agents/base/openai_responses_agent
-chmod +x init.sh deploy.sh
-./init.sh
+llama stack run ../../../run_llama_server.yaml
+```
+
+> **Keep this terminal open** – the server needs to keep running.  
+> You should see output indicating the server started on `http://localhost:8321`.
+
+Run the example:
+
+```bash
+uv run agents/base/openai_responses_agent/examples/execute_ai_service_locally.py
 ```
 
-This loads `.env`, validates variables, and copies `utils.py` into the agent package.
+---
+
+## Deployment on Red Hat OpenShift Cluster
 
-### Step 2: Build Image and Deploy
+Make deploy script executable:
+
+```bash
+chmod +x deploy.sh
+```
+
+Build image and deploy agent:
 
 ```bash
 ./deploy.sh
 ```
 
-This creates the API key secret, builds and pushes the image, and deploys the agent (Deployment, Service, Route).
+This will:
 
-### Step 3: Test the Agent
+- Create Kubernetes secret for API key
+- Build and push the Docker image
+- Deploy the agent to OpenShift
+- Create Service and Route
 
-Get the route host:
+Get the route URL:
 
 ```bash
 oc get route openai-responses-agent -o jsonpath='{.spec.host}'
@@ -115,9 +170,14 @@ Send a test request:
 ```bash
 curl -X POST https://<YOUR_ROUTE_URL>/chat \
   -H "Content-Type: application/json" \
-  -d '{"message": "How much does a Lenovo Laptop costs and what are the reviews?"}'
+  -d '{"message": "How much does a Lenovo Laptop cost and what are the reviews?"}'
 ```
 
-## References
+---
+
+## Agent-Specific Documentation
 
-- [OpenAI Python client](https://github.com/openai/openai-python) and [Responses API](https://platform.openai.com/docs/api-reference/responses/create)
+- [OpenAI Python client](https://github.com/openai/openai-python)
+- [OpenAI Responses API](https://platform.openai.com/docs/api-reference/responses/create)
+- [Ollama](https://ollama.com/)
+- [Ollama (Homebrew)](https://formulae.brew.sh/formula/ollama#default)
@@ -12,6 +12,10 @@ packages = [
 python = ">=3.12"
 python-dotenv = "^1.0.0"
 openai = ">=1.0.0"
+llama-stack = ">=0.5.0"
+setuptools = "69.5.1"
+milvus-lite = "2.5.1"
+pymilvus = "2.6.9"
 
 [tool.poetry.group.dev]
 optional = true

@@ -9,7 +9,7 @@ providers:
     - provider_id: milvus
       provider_type: inline::milvus
       config:
-        db_path: ${MILVUS_DB_PATH:-./milvus_data/milvus_lite.db}
+        db_path: ./milvus_data/milvus_lite.db
         persistence:
           namespace: vector_io::milvus
           backend: kv_default