Local Assistant WebUI Kit

This repository packages the documentation, prompt template, helper scripts, and launcher shortcuts needed to distribute {{PRODUCT_NAME}} as an Ollama-based assistant exposed through Open WebUI.

Contents

• QUICKSTART.md – concise setup instructions for installing Ollama, baking the {{ASSISTANT_NAME}} model, and running Open WebUI without Docker.
• Modelfile – template for ollama create; replace {{INSTRUCTIONS_PLACEHOLDER}} with the full prompt from prompts/instructions.md before building the model.
• prompts/ – system instructions template with placeholders for {{COMPANY_NAME}} to customize.
• knowledge/ – placeholder directory where users can drop PDFs/TXTs/CSVs prior to uploading through Open WebUI’s Knowledge feature.
• scripts/ – cross-platform helpers to pull the base model, bake instructions, and launch Open WebUI against the local Ollama service.
• launchers/ – OS-specific shortcuts that open http://localhost:8080 in the default browser.
• CONFIG.example.env – sample environment variables for default Ollama and WebUI endpoints.

Usage Workflow

1. Read and follow QUICKSTART.md to install prerequisites, customize the Modelfile, and create the {{ASSISTANT_NAME}} model from {{MODEL_TAG}}.
2. Update placeholders ({{PRODUCT_NAME}}, {{ASSISTANT_NAME}}, {{CONTACT_EMAIL}}, etc.) across the prompt and documentation before sharing with end users.
3. Use the helper scripts (create_model_*, start_webui_*) for a streamlined setup on macOS, Linux, or Windows.
4. Share the launchers with end users so they can open the hosted WebUI quickly once the server is running.

Network & Public Access

• Local network sharing: run Open WebUI with --host 0.0.0.0 and a fixed port (e.g., 8080), ensure the firewall allows inbound traffic, then have teammates visit http://<your-hostname>:8080. Keep Ollama bound to 127.0.0.1 so only the UI is network-exposed, and require authentication inside WebUI.
• Public exposure: front the WebUI with a reverse proxy (Nginx, Caddy, or cloud load balancer) that terminates TLS, enforces authentication, and optionally adds rate limits. If hosting externally, lock down the Ollama endpoint with a VPN or IP allow-list to prevent unauthorized access. Avoid publishing the raw Ollama port to the internet.

Customization & Maintenance

• Tailor prompts/instructions.md whenever {{COMPANY_NAME}} updates operating policies, then rebuild the model via the create script.
• Mirror edits to QUICKSTART.md or this README whenever processes change.
• Keep binaries, build artifacts, and large files out of version control; rely on the scripts and instructions instead of packaging compiled assets.

Support & Licensing

Document licensing separately (e.g., add a LICENSE file) and direct customers to {{CONTACT_EMAIL}} for assistance. Update copyright notices with {{YEAR}}.