fix: add localhost prefix to image names to prevent Docker registry auto-prefixing

a2a/exgentic_agent/.dockerignore

@@ -0,0 +1,38 @@
+# Git files
+.git
+.gitignore
+.gitattributes
+
+# Documentation
+*.md
+!README.md
+
+# IDE and editor files
+.vscode/
+.idea/
+*.swp
+*.swo
+*~
+
+# Python cache
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+
+# Virtual environments
+venv/
+env/
+ENV/
+
+# OS files
+.DS_Store
+Thumbs.db
+
+# Logs
+*.log
+
+# Temporary files
+*.tmp
+*.temp

a2a/exgentic_agent/.env.advanced

@@ -0,0 +1,4 @@
+OPENAI_API_KEY='{"valueFrom": {"secretKeyRef": {"name": "openai-secret", "key": "apikey"}}}'
+OPENAI_API_BASE='https://api.openai.com/v1' 
+AGENT_NAME=advanced_agent
+EXGENTIC_SET_AGENT_MODEL=openai/gpt-4o

a2a/exgentic_agent/.env.example

@@ -0,0 +1,3 @@
+OPENAI_API_KEY='{"valueFrom": {"secretKeyRef": {"name": "openai-secret", "key": "apikey"}}}'
+OPENAI_API_BASE='https://api.openai.com/v1' 
+EXGENTIC_SET_AGENT_MODEL=openai/gpt-4o

a2a/exgentic_agent/Dockerfile

@@ -0,0 +1,72 @@
+FROM ghcr.io/astral-sh/uv:python3.12-bookworm-slim
+
+# Build arguments
+ARG AGENT_NAME
+ARG RELEASE_VERSION="main"
+
+# Validate AGENT_NAME is provided
+RUN if [ -z "$AGENT_NAME" ]; then \
+        echo "ERROR: AGENT_NAME build argument is required"; \
+        echo "Usage: docker build --build-arg AGENT_NAME=<agent> -t exgentic-a2a-<agent> ."; \
+        exit 1; \
+    fi
+
+# Install system dependencies (git and git-lfs are needed for cloning)
+RUN apt-get update && apt-get install -y \
+    git \
+    git-lfs \
+    && rm -rf /var/lib/apt/lists/* \
+    && git lfs install
+
+# Configure git to handle large repositories and avoid HTTP/2 issues
+RUN git config --global http.postBuffer 524288000 && \
+    git config --global http.version HTTP/1.1 && \
+    git config --global core.compression 0
+
+# Set working directory
+WORKDIR /app
+
+# Clone the Exgentic repository
+RUN git clone https://github.com/Exgentic/exgentic.git /app/exgentic
+
+# Checkout the feature/mcp-command branch
+WORKDIR /app/exgentic
+RUN git checkout feature/mcp-command
+
+# Set UV_SYSTEM_PYTHON before any uv operations to ensure uv uses system Python
+ENV UV_SYSTEM_PYTHON=1
+
+# Install Exgentic and its dependencies using uv
+RUN uv pip install --system --no-cache -e .
+
+# Setup the agent - UV_SYSTEM_PYTHON is already set above
+RUN exgentic install --agent ${AGENT_NAME}
+
+# Copy entrypoint script
+WORKDIR /app
+COPY entrypoint.sh /app/entrypoint.sh
+RUN chmod +x /app/entrypoint.sh
+
+# Set environment variables
+ENV AGENT_NAME=${AGENT_NAME} \
+    HOST=0.0.0.0 \
+    PORT=8000 \
+    LOG_LEVEL=INFO \
+    PRODUCTION_MODE=True \
+    RELEASE_VERSION=${RELEASE_VERSION}
+
+# Create non-root user and set permissions
+RUN useradd -m -u 1001 exgentic && \
+    chown -R 1001:0 /app && \
+    chmod -R g+rwX /app
+
+USER 1001
+
+# Set HOME environment variable to /app
+ENV HOME=/app
+
+
+# Expose the default port
+EXPOSE 8000
+# Set the entrypoint
+ENTRYPOINT ["/app/entrypoint.sh"]

a2a/exgentic_agent/README.md

@@ -0,0 +1,275 @@
+# Exgentic A2A Agent Wrapper
+
+A Docker-based wrapper that runs [Exgentic](https://github.com/Exgentic/exgentic) agents using the A2A (Agent-to-Agent) protocol. This wrapper clones the Exgentic repository, installs a specific agent at build time, and exposes it via the A2A interface.  These agents are used for evaluation against the [Exgentic benchmarks mcp servers](mcp/README.md).  The test harness used to evaluate the agents against the benchmarks is found in the [workload-harness repo](https://github.com/kagenti/workload-harness/blob/main/exgentic_a2a_runner/README.md).
+
+## Overview
+
+This wrapper provides access to Exgentic agents through the A2A protocol. Each Docker image is built with a specific agent pre-installed, making it easy to deploy and run different agents in isolated environments.
+
+## Features
+
+- **Agent-Specific Images**: Each Docker image contains a single agent
+- **Build-Time Setup**: Agents are installed during image build for faster startup
+- **Flexible Configuration**: HOST and PORT configurable at runtime via environment variables
+- **Consistent Interface**: Follows A2A protocol standards
+- **Security**: Runs as non-root user (UID 1001)
+- **Production Ready**: Includes proper error handling and logging
+- **Build Script**: Convenient build.sh script with docker/podman auto-detection
+
+## Prerequisites
+
+- Docker or Podman installed on your system
+- Internet connection (for cloning repository and downloading agent data)
+- Sufficient disk space (agent data can be large)
+
+## Quick Start
+
+### Using the Build Script (Recommended)
+
+The easiest way to build an agent image:
+
+```bash
+cd a2a/exgentic_agent
+
+# Build an agent image
+./build.sh tool_calling
+```
+
+The script will:
+- Auto-detect docker or podman
+- Build the image with proper tagging
+- Provide colored output and progress information
+
+
+### Run the Agent
+
+```bash
+docker run -p 8000:8000 \
+  -e MCP_URL=http://host.containers.internal:8000/mcp \
+  -e EXGENTIC_SET_AGENT_MODEL='openai/gpt-4o' \
+  -e OPENAI_API_KEY \
+  -e OPENAI_API_BASE \
+  exgentic-a2a-tool_calling:latest
+
+```
+
+The agent will start on `http://0.0.0.0:8000`
+
+### Test the Agent
+
+```bash
+# Check if agent is running
+curl http://localhost:8000/health
+
+# List available capabilities
+curl http://localhost:8000/capabilities
+```
+
+## Build Script Usage
+
+The `build.sh` script provides a convenient way to build agent images:
+
+```bash
+# Basic usage
+./build.sh AGENT_NAME [--tag TAG] [--use-cache]
+
+# Examples
+./build.sh tool_calling                    # Build without cache (default)
+./build.sh tool_calling --tag v1.0.0       # Build v1.0.0 without cache
+./build.sh tool_calling --tag dev          # Build with 'dev' tag
+./build.sh tool_calling --use-cache        # Build with cache enabled
+./build.sh tool_calling --tag v1.0.0 --use-cache  # Build v1.0.0 with cache
+
+# Get help
+./build.sh --help
+```
+
+**Features:**
+- Automatically detects docker or podman
+- Colored output for better readability
+- Build summary with success/failure counts
+- Builds without cache by default for consistency
+- Optional cache usage with `--use-cache` flag
+
+## Configuration
+
+### Build Arguments
+
+| Argument | Required | Default | Description |
+|----------|----------|---------|-------------|
+| `AGENT_NAME` | Yes | - | The agent to install (e.g., tool_calling, agent1, agent2) |
+| `RELEASE_VERSION` | No | main | Version tag for tracking |
+
+### Runtime Environment Variables
+
+| Variable | Required | Default | Description |
+|----------|----------|---------|-------------|
+| `HOST` | No | 0.0.0.0 | Server host address |
+| `PORT` | No | 8000 | Server port |
+| `LOG_LEVEL` | No | INFO | Logging level (DEBUG, INFO, WARNING, ERROR) |
+| `AGENT_NAME` | No | (from build) | Agent name (set during build) |
+| `EXGENTIC_SET_*` | No | - | Runtime configuration parameters (see below) |
+
+### Runtime Configuration with --set Parameters
+
+You can pass runtime configuration parameters to the `exgentic a2a` command using environment variables with the `EXGENTIC_SET_` prefix. These will be converted to `--set` arguments.
+
+**Format**: `EXGENTIC_SET_<CATEGORY>_<PARAMETER>=<value>`
+- The `<CATEGORY>` will be separated from `<PARAMETER>` with a dot
+- The rest of the underscores in `<PARAMETER>` remain as underscores
+- Everything is converted to lowercase
+- Example: `EXGENTIC_SET_AGENT_MODEL` → `--set agent.model`
+
+**Common Parameters**:
+- `EXGENTIC_SET_AGENT_MODEL` - Set the agent model (e.g., `openai/gpt-4o`)
+- `EXGENTIC_SET_AGENT_MAX_STEPS` - Set maximum steps
+- `EXGENTIC_SET_AGENT_TEMPERATURE` - Set temperature
+- `EXGENTIC_SET_AGENT_TIMEOUT` - Set timeout
+- Any other agent-specific configuration parameter
+
+### Custom Configuration Examples
+
+
+**Set multiple runtime parameters:**
+```bash
+docker run -p 8000:8000 \
+  -e EXGENTIC_SET_AGENT_MODEL='openai/gpt-4o' \
+  -e EXGENTIC_SET_AGENT_MAX_STEPS='50' \
+  -e EXGENTIC_SET_AGENT_TEMPERATURE='0.7' \
+  exgentic-a2a-tool_calling:latest
+```
+
+### API Credentials
+
+When using external models, you need to provide API credentials as environment variables:
+
+| Variable | Required | Description |
+|----------|----------|-------------|
+| `OPENAI_API_KEY` | Yes (for OpenAI models) | Your OpenAI API key |
+| `OPENAI_API_BASE` | No | Custom API base URL (if using a proxy or alternative endpoint) |
+
+**Example with OpenAI credentials:**
+```bash
+docker run -p 8000:8000 \
+  -e OPENAI_API_KEY='your-api-key-here' \
+  -e EXGENTIC_SET_AGENT_MODEL='openai/gpt-4o' \
+  exgentic-a2a-tool_calling:latest
+```
+
+**Example with custom API base:**
+```bash
+docker run -p 8000:8000 \
+  -e OPENAI_API_KEY='your-api-key-here' \
+  -e OPENAI_API_BASE='https://custom-endpoint.example.com/v1' \
+  -e EXGENTIC_SET_AGENT_MODEL='openai/gpt-4o' \
+  exgentic-a2a-tool_calling:latest
+```
+
+## Build Process
+
+```
+1. Clone Exgentic repository (HTTPS)
+   ↓
+2. Checkout feature/mcp-command branch
+   ↓
+3. Install Exgentic and dependencies
+   ↓
+4. Run: exgentic install --agent $AGENT_NAME
+   ↓
+5. Configure entrypoint and permissions
+   ↓
+6. Create image: exgentic-a2a-{agent}:latest
+```
+
+## Runtime Process
+
+```
+1. Container starts with entrypoint.sh
+   ↓
+2. Read environment variables (HOST, PORT)
+   ↓
+3. Execute: exgentic a2a --agent $AGENT_NAME --host $HOST --port $PORT
+   ↓
+4. A2A agent listens on configured HOST:PORT
+```
+
+## Image Naming Convention
+
+Images follow the pattern: `exgentic-a2a-{agent}:latest`
+
+Examples:
+- `exgentic-a2a-tool_calling:latest`
+- `exgentic-a2a-agent1:latest`
+- `exgentic-a2a-agent2:latest`
+
+
+## Repository Information
+
+- **Source**: https://github.com/Exgentic/exgentic.git
+- **Branch**: feature/mcp-command
+- **Protocol**: HTTPS (public repository)
+
+## A2A Protocol
+
+This wrapper implements the Agent-to-Agent (A2A) protocol, which allows agents to communicate and collaborate programmatically.
+
+### Key Features
+
+- **Agent Discovery**: Agents expose their capabilities via A2A
+- **Type Safety**: Strong typing for parameters and returns
+- **Documentation**: Built-in documentation for each capability
+- **Error Handling**: Standardized error responses
+- **Transport**: HTTP transport with streamable support
+
+## Security Considerations
+
+- Runs as non-root user (UID 1001)
+- No SSH keys or secrets in image
+- Public repository access only
+- Minimal attack surface
+- Production-ready configuration
+
+## Performance Notes
+
+- **Build Time**: 5-15 minutes depending on agent size
+- **Image Size**: Varies by agent (typically 1-5 GB)
+- **Startup Time**: Fast (agent already installed)
+- **Memory**: Depends on agent requirements
+
+## Files in This Directory
+
+- `Dockerfile` - Multi-stage build configuration
+- `entrypoint.sh` - Container startup script
+- `build.sh` - Convenient build script with auto-detection
+- `.dockerignore` - Files to exclude from build context
+- `.env.example` - Basic environment variable template
+- `.env.advanced` - Advanced configuration example
+- `README.md` - This file
+
+## Contributing
+
+When adding support for new agents:
+
+1. Verify the agent exists in the Exgentic repository
+2. Test the build process using `./build.sh agent_name`
+3. Document any special requirements
+4. Update this README with examples
+
+## License
+
+See the repository's LICENSE file for details.
+
+## Support
+
+For issues related to:
+- **This A2A wrapper**: Open an issue in the agent-examples repository
+- **Exgentic agents**: Refer to the Exgentic repository
+- **A2A protocol**: See the Agent-to-Agent protocol documentation
+
+## Related Resources
+
+- [Exgentic Repository](https://github.com/Exgentic/exgentic)
+- [Agent-to-Agent Protocol](https://github.com/Exgentic/exgentic)
+- [Docker Documentation](https://docs.docker.com/)
+- [Podman Documentation](https://podman.io/)
+- [Other A2A Agents](../README.md)