M.I.M.I.R - Multi-agent Intelligent Memory & Insight Repository

# Mimir Open-WebUI Custom Image This directory contains a custom Dockerfile that extends the official Open-WebUI image with: 1. **Neo4j Python driver** pre-installed 2. **Mimir pipelines** pre-packaged and ready to enable 3. **Required dependencies** (aiohttp, pydantic) pre-installed ## Purpose The Mimir orchestration pipeline requires the `neo4j` Python package to interact with the Neo4j graph database. By baking this dependency and the pipeline files into the image at build time, we provide a seamless out-of-the-box experience. ## Pre-Packaged Pipelines The following Mimir pipelines are automatically copied into the container: 1. **mimir_orchestrator.py** - Multi-agent orchestration (Ecko → PM → Workers → QC) 2. **mimir_rag_auto.py** - RAG with semantic search context enrichment 3. **mimir_file_browser.py** - File browsing actions 4. **mimir_tools_wrapper.py** - Command shortcuts (`/list_folders`, `/search`, etc.) **Location in container**: `/app/backend/data/pipelines/` **Status**: Pre-installed but disabled by default. Must be manually enabled via Admin Panel. ## Dockerfile ```dockerfile FROM ghcr.io/open-webui/open-webui:main USER root # Install Python dependencies for Mimir pipelines RUN pip install --no-cache-dir \ neo4j \ aiohttp \ pydantic # Create pipelines directory RUN mkdir -p /app/backend/data/pipelines # Copy Mimir pipelines into the container COPY pipelines/mimir_orchestrator.py /app/backend/data/pipelines/ COPY pipelines/mimir_rag_auto.py /app/backend/data/pipelines/ COPY pipelines/mimir_file_browser.py /app/backend/data/pipelines/ COPY pipelines/mimir_tools_wrapper.py /app/backend/data/pipelines/ # Copy init script to auto-enable pipelines COPY docker/open-webui/init-pipelines.sh /app/init-pipelines.sh RUN chmod +x /app/init-pipelines.sh # Set ownership RUN chown -R root:root /app/backend/data/pipelines USER root ``` This Dockerfile: 1. Extends the official Open-WebUI base image 2. Switches to root user for package installation 3. Installs required Python packages with no cache to keep image size small 4. Creates the pipelines directory 5. Copies all 4 Mimir pipeline files into the container 6. Copies an init script for pipeline verification 7. Sets proper ownership ## Building The image is automatically built via docker-compose: ```bash # Build just the open-webui service docker compose build open-webui # Or use the npm script npm run webui:build # Or rebuild everything npm run build:docker ``` The image is tagged as: - `mimir-open-webui:1.0.0` (version from package.json) - `mimir-open-webui:latest` ## Enabling Pipelines After building and starting the container, pipelines must be manually enabled: ### Via Admin Panel (Recommended) 1. Access Open-WebUI at http://localhost:3000 2. Click your username (bottom-left) → **"Admin Panel"** 3. Navigate to **"Pipelines"** tab 4. Toggle each Mimir pipeline to enable it: - ✅ Mimir Multi-Agent Orchestrator - ✅ Mimir RAG Auto - ✅ Mimir File Browser - ✅ Mimir Tools Wrapper **See**: [PIPELINE_CONFIGURATION.md](../../docs/guides/PIPELINE_CONFIGURATION.md) for detailed instructions. ## Verification After building and starting the container, verify the pipelines are copied: ```bash # List pipeline files docker exec -it mimir-open-webui ls -la /app/backend/data/pipelines/ # Should show: # mimir_orchestrator.py # mimir_rag_auto.py # mimir_file_browser.py # mimir_tools_wrapper.py ``` Verify dependencies are installed: ```bash # Check Neo4j driver docker exec -it mimir-open-webui pip show neo4j # Check all dependencies docker exec -it mimir-open-webui pip list | grep -E "neo4j|aiohttp|pydantic" ``` You should see: ``` Name: neo4j Version: 6.0.3 Summary: Neo4j Bolt driver for Python Name: aiohttp Version: 3.9.1 Summary: Async http client/server framework Name: pydantic Version: 2.5.0 Summary: Data validation using Python type hints ``` ## Updating Pipelines To update pipelines after making changes: ```bash # 1. Edit pipeline files in pipelines/ directory nano pipelines/mimir_orchestrator.py # 2. Rebuild the image (no cache to ensure fresh copy) docker compose build open-webui --no-cache # 3. Restart the container docker compose up -d --force-recreate open-webui ``` ## Updating Dependencies To add more Python packages: 1. Edit the Dockerfile and add additional `pip install` commands 2. Rebuild: `docker compose build open-webui` 3. Restart: `docker compose up -d --no-deps open-webui` Or create a `requirements.txt` file and install from it: ```dockerfile FROM ghcr.io/open-webui/open-webui:main USER root COPY requirements.txt /tmp/ RUN pip install --no-cache-dir -r /tmp/requirements.txt && rm /tmp/requirements.txt ``` ## Troubleshooting ### Pipelines Not Showing Up ```bash # 1. Verify files are in container docker exec -it mimir-open-webui ls -la /app/backend/data/pipelines/ # 2. Check Open-WebUI logs docker compose logs open-webui | grep -i pipeline # 3. Rebuild with no cache docker compose build open-webui --no-cache docker compose up -d --force-recreate open-webui ``` ### Pipeline Fails to Enable ```bash # 1. Check for Python syntax errors docker exec -it mimir-open-webui python3 -m py_compile \ /app/backend/data/pipelines/mimir_orchestrator.py # 2. Check dependencies docker exec -it mimir-open-webui pip list | grep -E "neo4j|aiohttp|pydantic" # 3. Check logs for errors docker compose logs open-webui | tail -50 ``` ## Related Documentation - **[PIPELINE_CONFIGURATION.md](../../docs/guides/PIPELINE_CONFIGURATION.md)** - Complete pipeline setup guide - **[LLM_PROVIDER_GUIDE.md](../../docs/guides/LLM_PROVIDER_GUIDE.md)** - Switching LLM providers - **[QUICKSTART.md](../../docs/getting-started/QUICKSTART.md)** - Initial setup