add ollama service

[jasielmacedo]

Description

This PR bundles Ollama with the application, adds auto-start functionality, implements a model management UI button, and sets up automated binary downloads for contributors. Users no longer need to manually install Ollama separately.

Type of Change

• New feature (non-breaking change that adds functionality)
• Code quality improvement (refactoring, formatting, etc.)
• Documentation update

Changes Made

Core Features

• Bundled Ollama v0.12.9: Application now includes Ollama executables for Windows (x64) and macOS (Intel + Apple Silicon)
• Auto-start on Launch: Ollama server automatically starts when the application launches (no manual ollama serve required)
• Model Manager UI Button: Added visible button in NavigationBar to open Model Manager (previously only accessible via Ctrl+M)
• Automatic Binary Downloads: Created scripts/download-ollama.js to download Ollama binaries automatically during npm install

Implementation Details

• Modified src/main/index.ts:
  • Added ollamaService.ensureRunning() in app.whenReady() callback
  • Added cleanup on quit (ollamaService.stop())
  • Made callback async to support auto-start
• Updated src/main/services/ollama.ts:
  • Added getBundledOllamaPath() method to locate platform-specific Ollama executable
  • Updated start() method to use bundled executable instead of system PATH
  • Set environment variables for Windows to find required DLLs
  • Fixed 4 ESLint errors by prefixing unused catch variables with underscore
• Modified src/renderer/components/Browser/NavigationBar.tsx:
  • Added Model Manager button with CPU/chip icon
  • Integrated with useModelStore for state management

Build & Distribution

• Created electron-builder.json with configuration to bundle Ollama in production builds
• Added extraResources to include resources/bin/ in distribution
• Updated .gitignore to exclude resources/bin/ directory (~1.8GB)

Developer Experience

• Created scripts/download-ollama.js:
  • Downloads Ollama v0.12.9 for Windows and macOS from GitHub releases
  • Extracts to resources/bin/win32/ and resources/bin/darwin/
  • Shows download progress
  • Skips if binaries already exist
• Added npm scripts:
  • setup:ollama - Manually download Ollama binaries
  • postinstall - Automatically runs after npm install
• Created scripts/README.md with usage and troubleshooting documentation

Documentation

• Updated README.md:
  • Documented bundled Ollama version (v0.12.9, released Nov 1, 2025)
  • Added "Using AI Features" section
  • Added "Bundled Ollama Version" section with update instructions
  • Updated Getting Started to reflect automatic binary download
  • Updated Contributing guidelines
• Updated TECH_BRIEFING.md:
  • Updated LLM Integration section with bundled version info
  • Updated file structure to show resources/bin/ directory
  • Updated code examples to show bundled executable usage
  • Updated build documentation
  • Updated environment requirements

Code Quality

• Updated eslint.config.js:
  • Added caughtErrorsIgnorePattern: '^_' to ignore underscore-prefixed catch variables
• Fixed ESLint errors in ollama.ts (4 unused variable warnings)

Testing

• Tested locally in development mode
• Tested production build
• Manually tested affected features

Test Results

• ✅ Build completes successfully (npm run build)
• ✅ Ollama auto-starts when app launches (tested on Windows)
• ✅ Model Manager button visible and functional in NavigationBar
• ✅ Download script works correctly (npm run setup:ollama)
• ✅ Binaries are properly excluded from git
• ✅ ESLint passes with no errors for modified files
• ✅ Platform detection works for both Windows and macOS paths

Screenshots

Model Manager Button in NavigationBar:

• New button added between Bookmarks and AI Chat buttons
• Icon: CPU/chip design (matching other navigation controls)
• Tooltip: "Model Manager (Ctrl+M)"

Download Script Output:

╔═══════════════════════════════════════════════════════════╗
║        Ollama Binary Download Script                     ║
║        Version: v0.12.9                                   ║
╚═══════════════════════════════════════════════════════════╝

[1/2] Windows (x64)
────────────────────────────────────────────────────────────
  Downloading from: https://github.com/ollama/ollama/releases/...
  Progress: 100%
  Extracting to: resources/bin/win32
  ✓ Complete

[2/2] macOS (Universal)
────────────────────────────────────────────────────────────
  ✓ Already downloaded, skipping...

╔═══════════════════════════════════════════════════════════╗
║              Download Complete! ✓                         ║
╚═══════════════════════════════════════════════════════════╝

Checklist

• My code follows the project's code style (ESLint and Prettier)
• I have performed a self-review of my code
• I have commented my code where necessary
• My changes generate no new warnings or errors
• I have tested my changes locally

Additional Notes

Bundle Size Impact

• Git Repository: No change (binaries excluded via .gitignore)
• Final Installer: ~1.8GB (Windows includes CUDA 12/13, ROCm libraries for GPU acceleration)
• macOS Bundle: ~46MB (smaller, no CUDA required)

Breaking Changes

None. This is a fully backward-compatible enhancement. The app now works out-of-the-box without requiring users to install Ollama separately.

Future Improvements

• Consider adding Linux support (currently Windows + macOS only)
• Could add automatic update checks for newer Ollama versions
• May want to add option to use system-installed Ollama if user prefers

Migration Notes for Contributors

After pulling this PR:

1. Run npm install (this will download Ollama binaries automatically)
2. If npm install was already run before this PR, manually run: npm run setup:ollama
3. The binaries are stored in resources/bin/ (excluded from git, ~1.8GB)