Provide three very simple AI examples

ai/agent/.env.sample

@@ -0,0 +1,13 @@
+#INFERENCE_BASE_URL=http://0.0.0.0:11434/v1
+#INFERENCE_API_KEY="dummy"
+#INFERENCE_MODEL_NAME="openai/qwen3:4b"
+
+#INFERENCE_BASE_URL=https://api.perplexity.ai/v1
+#INFERENCE_API_KEY="REDACTED"
+#INFERENCE_MODEL_NAME="perplexity/sonar-pro"
+
+INFERENCE_BASE_URL=https://eu-de.ml.cloud.ibm.com
+INFERENCE_API_KEY="REDACTED"
+INFERENCE_MODEL_NAME="watsonx/meta-llama/llama-3-3-70b-instruct"
+INFERENCE_PROJECT_ID="REDACTED"
+

ai/agent/.gitignore

@@ -0,0 +1 @@
+.env

ai/agent/README.md

@@ -0,0 +1,95 @@
+Agent Overview
+--------------
+
+An agent is a small, purpose-built service that performs autonomous tasks by combining conversational logic, tool wrappers, and connectors to external services. Agents typically:
+
+- **Purpose**: Automate workflows and respond to user inputs or events.
+- **Components**: Orchestration logic, task definitions, tool adapters (APIs, search, databases), and an optional lightweight UI.
+
+![](./images/architecture.png)
+
+Why IBM Cloud Code Engine
+--------------
+
+[IBM Cloud Code Engine](https://www.ibm.com/products/code-engine) is a great fit for containerized agents because it provides:
+
+- **Serverless containers**: Deploy container images without managing infrastructure.
+- **Automatic scaling**: Scale to zero when idle and scale up on demand.
+- **Pay-per-use pricing**: Cost-efficient for intermittent workloads common to agents.
+- **Simple deployment**: Integrates with container registries and CI/CD pipelines.
+- **Managed endpoint**: Provides a secure http endpoint with a managed certificate.
+
+A very simple Agent
+--------------
+
+
+The agent implementation in this folder demonstrates a compact, production-minded agent pattern. It contains orchestration, task definitions, and tool wrappers that together implement a conversational assistant with tooling capabilities (search, task execution, and basic web UI).
+
+- **Agent entrypoint**: [./src/main.py](./src/main.py) — starts the agent process and HTTP endpoints.
+- **Agent orchestration**: [./src/agents.py](./src/agents.py) — coordinates agent behavior and message routing.
+- **Tasks**: [./src/tasks.py](./src/tasks.py) — concrete task implementations the agent can run.
+- **Tools / adapters**: [./src/tools.py](./src/tools.py) — wrappers for external APIs or search capabilities.
+- **Utilities**: [./src/utils.py](./src/utils.py) — helpers and common utilities.
+- **Frontend**: [./src/frontend/landing_page.py](./src/frontend/landing_page.py) — lightweight landing page for human interaction (optional).
+- **Example payload**: [./payload/payload.json](./agent/payload/payload.json) — sample input the agent can accept.
+
+Conceptually this agent acts like a Crew AI / ACP-style assistant: it receives a payload, enriches context (via tools/search), executes defined tasks, and returns results. The implementation is intentionally modular so you can swap tool implementations or extend task behaviors.
+
+
+
+
+Deploy the Agent to IBM Cloud Code Engine
+--------------
+
+**Prerequisites**
+
+The Agent requires an nferencing backend (an LLM or similar service) to operate. Configure the backend endpoint and credentials via environment variables. Configure any OpenAI compatible API endpoint such as [OpenAI](https://platform.openai.com/) or [IBM watsonx.ai](https://www.ibm.com/products/watsonx-ai)
+
+Copy the template
+```bash
+# If you have a .env.template
+cp .env.template .env
+```
+Edit the parameter
+```bash
+INFERENCE_BASE_URL=https://eu-de.ml.cloud.ibm.com
+INFERENCE_API_KEY="<YOUR API KEY>"
+INFERENCE_MODEL_NAME="watsonx/meta-llama/llama-3-3-70b-instruct"
+INFERENCE_PROJECT_ID="<YOUR PROJECT>"
+```
+
+**Deploy**
+
+1. Log in to IBM Cloud (choose the method that fits your account):
+
+```bash
+# Single-sign-on login (interactive)
+ibmcloud login --sso
+
+# Or with API key (non-interactive):
+ibmcloud login --apikey "$IBMCLOUD_APIKEY" -r REGION -g RESOURCE_GROUP
+```
+
+2. Change into the agent folder and run the provided deploy script:
+
+```bash
+NAME_PREFIX=ce-agent REGION=eu-de ./deploy.sh
+```
+
+3. The [deploy script](./agent/deploy.sh) builds and/or pushes the container and deploys it to IBM Cloud Code Engine. 
+
+
+4. Follow the script output for the agents URL and an example http request.
+
+
+What's next 
+---------
+
+- Modify the tasks in [./src/tasks.py](./src/tasks.py) and register them with the orchestration logic in [./src/agents.py](./src/agents.py).
+- Add your own tools by editing [./src/tools.py](./src/tools.py).
+- Deploy another agent and add it to the Crew [./src/main.py](./src/main.py)
+- [Deploy a MCP Server](../mcp_server_fastmcp/README.md) and make it accessible to your agent(s)
+
+Follow the [IBM Cloud Code Engine documentation](https://cloud.ibm.com/docs/codeengine?topic=codeengine-getting-started) to learn about more features and functions.
+
+![](./images/architecture2.png)

ai/agent/deploy.sh

@@ -0,0 +1,138 @@
+#!/bin/bash
+
+# ==============================
+# ENVIRONMENT SETUP
+# ==============================
+
+source .env
+
+REGION="${REGION:=eu-de}"
+NAME_PREFIX="${NAME_PREFIX:=ce-agent}"
+
+ce_project_name=${PROJECT_NAME:-${NAME_PREFIX}-project}
+resource_group_name=${NAME_PREFIX}-rg
+
+# ==============================
+# COMMON FUNCTIONS
+# ==============================
+
+SCRIPT_DIR=$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )
+source "${SCRIPT_DIR}/../common.sh"
+
+# Clean up previous run
+function clean() {
+    (
+        ibmcloud ce project select --name ${ce_project_name} --quiet 2>/dev/null
+        if [ $? == 0 ]; then
+            ibmcloud ce project delete --name ${ce_project_name} --force --hard --no-wait 2>/dev/null
+        fi
+
+        ibmcloud resource group $resource_group_name --quiet 2>/dev/null
+        if [[ $? == 0 ]]; then
+            COUNTER=0
+            # some resources (e.g. boot volumes) are deleted with some delay. Hence, the script waits before exiting with an error
+            while (( "$(ibmcloud resource service-instances --type all -g $resource_group_name --location $REGION --output json | jq -r '. | length')" > 0 )); do
+                sleep 5
+                COUNTER=$((COUNTER + 1))
+                if ((COUNTER > 3)); then
+                    print_error "Cleanup failed! Please make sure to delete remaining resources manually to avoid unwanted charges."
+                    ibmcloud resource service-instances --type all -g $resource_group_name --location $REGION
+                    exit 1
+                fi
+            done
+        fi
+
+	ibmcloud resource group-delete $resource_group_name --force 2>/dev/null
+    )
+}
+
+if [[ "$1" == "clean" ]]; then
+    print_msg "\nCleaning up the created IBM Cloud resources ..."
+    clean
+    print_success "\n==========================================\n DONE\n==========================================\n"
+    exit 0
+fi
+
+# ==============================
+# MAIN SCRIPT FLOW
+# ==============================
+
+# Ensure that latest versions of used IBM Cloud ClI is installed
+print_msg "\nPulling latest IBM Cloud CLI release ..."
+ibmcloud update --force
+
+# Ensure that latest versions of used IBM Cloud CLI plugins are installed
+print_msg "\nInstalling required IBM Cloud CLI plugins ..."
+ensure_plugin_is_up_to_date code-engine
+
+print_msg "\n======================================================"
+print_msg " Setting up \"Code Engine AI Agent\" sample"
+print_msg "======================================================\n"
+
+target_region $REGION
+
+#
+# Create the resource group, if it does not exist
+ibmcloud resource group $resource_group_name --quiet
+if [ $? != 0 ]; then
+    print_msg "\nCreating resource group '$resource_group_name' ..."
+    ibmcloud resource group-create $resource_group_name
+fi
+target_resource_group $resource_group_name
+
+if ! does_instance_exist codeengine "$ce_project_name"; then
+    print_msg "\nCreating the Code Engine project '$ce_project_name' ..."
+    ibmcloud ce project create --name $ce_project_name
+    if [ $? -ne 0 ]; then
+        print_error "Code Engine project creation failed!"
+        abortScript
+    fi
+else
+    print_msg "\nSelecting the Code Engine project '$ce_project_name' ..."
+    ibmcloud ce project select --name $ce_project_name
+fi
+project_guid=$(ibmcloud ce project get --name $ce_project_name --output json | jq -r '.guid')
+
+print_msg "\nCreating code engine secret for the agents with parameters from .env file"
+ibmcloud ce secret create --name agent-secret --from-literal INFERENCE_BASE_URL="$INFERENCE_BASE_URL" \
+--from-literal INFERENCE_API_KEY="$INFERENCE_API_KEY" \
+--from-literal INFERENCE_MODEL_NAME="$INFERENCE_MODEL_NAME" \
+--from-literal INFERENCE_PROJECT_ID="$INFERENCE_PROJECT_ID"
+
+print_msg "\nCreating the agent as a code engine application"
+ibmcloud ce app create --name agent --env-from-secret agent-secret --build-source ./src --build-strategy dockerfile  --cpu 1 --memory 4G -p 8080  --wait-timeout 600 --min-scale 0 --visibility public
+
+while true; do
+    ibmcloud ce app list
+    not_ready_apps=$(ibmcloud ce app list | grep -e "agent" | grep "Not Ready")
+    if [  "$not_ready_apps" == "" ]; then
+      break # all apps are ready
+    fi
+    print_msg "\nWaiting for all applications to be ready (sleep 15s)..."
+    sleep 15
+done
+
+print_msg "\n"
+print_msg "\nAgent application is ready."
+print_msg "\n"
+
+url=$(ibmcloud ce app get --name agent -o json | jq -r '.status.url')
+print_msg "\nAgent is reachable under the following url:"
+echo $url
+
+print_msg "\n"
+print_msg "\nList the agents:"
+curl -s -X GET $url/agents | jq
+
+
+print_msg "\n"
+print_msg "\nCalling the agent with the following payload"
+cat payload/payload.json | jq
+
+curl -s -X POST $url/runs -H "Content-Type: application/json" -d @payload/payload.json | jq
+
+print_success "\n=========================================="
+print_success " SUCCESS"
+print_success "==========================================\n"
+print_msg "\n"
+print_msg "\nrun ./deploy.sh clean to remove all created resources"

ai/agent/payload/payload.json

@@ -0,0 +1 @@
+{"agent_name": "City_Expert", "input": [{"role": "User", "parts": [{"content": "{\"city\": \"Stuttgart\", \"country\": \"Germany\", \"currency\": \"EUR\"}"}]}]}

ai/agent/src/.dockerignore

@@ -0,0 +1,33 @@
+__pycache__/
+.my_cache_dir
+.pytest_cache
+.ruff_cache
+*.pyc
+*.pyo
+*.pyd
+*.db
+*.sqlite3
+
+.env
+.venv
+*.env
+
+**/.env
+**/.venv
+**/*.env
+
+.git
+.gitignore
+Dockerfile
+.dockerignore
+
+*.log
+*.bak
+*.swp
+
+.vscode/
+.idea/
+
+# OS-specific
+.DS_Store
+Thumbs.db

ai/agent/src/.gitignore

@@ -0,0 +1,13 @@
+# Python-generated files
+__pycache__/
+*.py[oc]
+build/
+dist/
+wheels/
+*.egg-info
+.mypy_cache
+.coverage
+.python-version
+
+# Virtual environments
+.venv

ai/agent/src/Dockerfile

@@ -0,0 +1,53 @@
+#################################################
+# Build stage
+#################################################
+
+FROM registry.access.redhat.com/ubi9/python-312:latest AS build-stage
+
+USER 1001
+WORKDIR $HOME
+
+ENV PATH="$HOME/.venv/bin:$PATH"
+
+# Install uv.
+COPY --from=ghcr.io/astral-sh/uv:latest /uv /uvx /bin/
+
+# Copy the application into the container.
+COPY . .
+
+# Install the application dependencies.
+RUN uv sync --frozen --no-cache
+
+USER root
+
+# https://docs.trychroma.com/updates/troubleshooting#sqlite
+RUN uv add pysqlite3-binary
+RUN sed -i '1i # ruff: noqa: E402,I001\n__import__("pysqlite3")\nimport sys\nsys.modules["sqlite3"] = sys.modules.pop("pysqlite3")' main.py
+
+# Linting
+RUN ruff check . --no-cache
+
+# Reinstall without dev dependencies
+RUN uv sync --frozen --no-dev
+
+USER 1001
+
+#################################################
+# Production stage
+#################################################
+
+FROM registry.access.redhat.com/ubi9/python-312:latest
+
+USER 1001
+WORKDIR $HOME
+
+ENV PATH="$HOME/.venv/bin:$PATH"
+ENV MODE=${MODE:-production}
+ENV LOG_LEVEL=${LOG_LEVEL:-DEBUG}
+ENV PORT=${PORT:-5630}
+ENV WORKERS=${WORKERS:-1}
+
+COPY --from=build-stage $HOME/ .
+
+# Run the application.
+CMD uvicorn main:app --host 0.0.0.0 --port ${PORT} --workers ${WORKERS} --log-level ${LOG_LEVEL,,} --timeout-keep-alive 50  --header server:acp

ai/agent/src/README.md

@@ -0,0 +1,9 @@
+# City Expert Agent
+
+```sh
+source .venv/bin/activate
+```
+
+```sh
+uv run main.py
+```

ai/agent/src/agents.py

@@ -0,0 +1,30 @@
+import os
+
+from crewai import LLM, Agent
+from dotenv import load_dotenv
+
+from tools import duckduckgosearch_tool, scrape_tool
+
+load_dotenv()
+
+llm = LLM(
+    model=os.getenv("INFERENCE_MODEL_NAME"),
+    base_url=os.getenv("INFERENCE_BASE_URL"),
+    api_key=os.getenv("INFERENCE_API_KEY", ""),
+    project_id=os.getenv("INFERENCE_PROJECT_ID"),
+    max_tokens=8192,
+    temperature=0,
+)
+
+city_local_expert_agent = Agent(
+    role="Local City Expert",
+    goal="Provide the BEST insights about the selected city",
+    backstory="A knowledgeable local guide with extensive information about the city, it's attractions and customs",
+    tools=[
+        scrape_tool,
+        duckduckgosearch_tool,
+    ],
+    reasoning=True,
+    verbose=True,
+    llm=llm,
+)