OmniContext VS Code Extension

Intelligent code context engine with pre-fetch caching and automatic context injection for AI coding assistants

Features

Pre-Fetch Context Caching

The extension intelligently tracks your IDE activity and pre-fetches relevant code context before you even ask. This dramatically reduces response time when working with AI assistants.

Screenshot Coming Soon: Sidebar showing cache statistics with real-time hit rate, hits, misses, and cache size display

How it works:

• Monitors file opens, cursor movements, and text edits
• Extracts symbols at cursor position using VS Code's language features
• Pre-fetches relevant context from the OmniContext daemon
• Caches results with configurable TTL (time-to-live)
• Automatically injects cached context into AI chat requests

Cache Indicators:

• [cached] - Context retrieved from cache (instant response)
• [fresh search] - Context fetched from index (first time or cache expired)

Screenshots Coming Soon:

• AI chat response showing cached indicator
• AI chat response showing fresh search indicator

Real-Time Cache Statistics

Monitor pre-fetch performance directly in the VS Code sidebar:

• Hit Rate - Percentage of requests served from cache
• Cache Hits - Number of successful cache retrievals
• Cache Misses - Number of fresh searches performed
• Cache Size - Current entries / Maximum capacity

Screenshot Coming Soon: Sidebar cache statistics section showing active status with performance metrics

Automatic Context Injection

When you interact with AI assistants (GitHub Copilot Chat, etc.), the extension automatically:

1. Detects your current file and cursor position
2. Checks the pre-fetch cache for relevant context
3. Injects context into the AI request (invisible to you)
4. Shows cache hit/miss indicator in the response

Flexible Configuration

Fine-tune pre-fetch behavior to match your workflow:

• Enable/disable pre-fetch functionality
• Adjust cache size (10-1000 entries)
• Configure cache TTL (60-3600 seconds)
• Set debounce delay for IDE events (50-1000ms)

Installation

1. Install the extension from VS Code Marketplace (coming soon) or build from source
2. Install OmniContext daemon: See main installation guide
3. Open a workspace - the extension will auto-start the daemon and begin indexing

Configuration

Pre-Fetch Settings

Access via VS Code Settings (Ctrl+, or Cmd+,) and search for "OmniContext"

Screenshot Coming Soon: VS Code settings page showing all OmniContext pre-fetch configuration options

omnicontext.prefetch.enabled

• Type: Boolean
• Default: true
• Description: Enable pre-fetch functionality to cache context based on IDE events (file open, cursor movement, text edits)

omnicontext.prefetch.cacheSize

• Type: Number
• Default: 100
• Range: 10-1000
• Description: Maximum number of entries in the pre-fetch cache. Higher values use more memory but improve hit rate.

Recommendations:

• Small projects (<1000 files): 50-100 entries
• Medium projects (1000-5000 files): 100-200 entries
• Large projects (>5000 files): 200-500 entries

omnicontext.prefetch.cacheTtlSeconds

• Type: Number
• Default: 300 (5 minutes)
• Range: 60-3600 seconds
• Description: Time-to-live for cached entries. Entries older than this are evicted.

Recommendations:

• Fast-changing codebases: 180-300 seconds (3-5 minutes)
• Stable codebases: 600-1800 seconds (10-30 minutes)
• Read-only exploration: 1800-3600 seconds (30-60 minutes)

omnicontext.prefetch.debounceMs

• Type: Number
• Default: 200
• Range: 50-1000 milliseconds
• Description: Debounce delay for IDE events (cursor movement, text edits) to avoid excessive pre-fetch requests

Recommendations:

• Fast typists: 300-500ms (reduce noise)
• Slow/deliberate navigation: 100-200ms (more responsive)
• High-latency systems: 500-1000ms (reduce load)

Other Settings

omnicontext.autoStartDaemon

• Type: Boolean
• Default: true
• Description: Automatically start the OmniContext daemon when opening a workspace

omnicontext.contextInjection

• Type: Boolean
• Default: true
• Description: Enable automatic context injection into AI chat requests

omnicontext.tokenBudget

• Type: Number
• Default: 8192
• Description: Token budget for context injection (limits context size)

Usage

Sidebar Controls

Open the OmniContext sidebar from the Activity Bar (left side) to access:

Pre-Fetch Cache Section:

• View real-time cache statistics
• Enable/disable pre-fetch functionality
• Clear cache manually
• Monitor cache status (Active/Disabled/Offline)

Context Control Section:

• Toggle automatic context injection
• View daemon status
• Start/stop daemon manually

Commands

Access via Command Palette (Ctrl+Shift+P or Cmd+Shift+P):

• OmniContext: Index Workspace - Manually trigger workspace indexing
• OmniContext: Search Code - Search codebase with semantic search
• OmniContext: Show Status - Display index statistics
• OmniContext: Start Daemon - Start the background daemon
• OmniContext: Stop Daemon - Stop the background daemon
• OmniContext: Toggle Context Injection - Enable/disable automatic context injection
• OmniContext: Refresh Sidebar - Refresh sidebar statistics

Cache Hit Rate Expectations

Good Performance (>60% hit rate)

• You're working in a focused area of the codebase
• Cache size and TTL are well-tuned for your workflow
• Daemon is running smoothly

Fair Performance (30-60% hit rate)

• You're exploring different parts of the codebase
• Cache size might be too small for your project
• Consider increasing cacheSize or cacheTtlSeconds

Poor Performance (<30% hit rate)

• You're jumping between many different files rapidly
• Cache size is too small or TTL is too short
• Daemon might be experiencing issues

Optimization tips:

1. Increase cacheSize to 200-500 for large projects
2. Increase cacheTtlSeconds to 600-1800 for stable codebases
3. Check daemon logs for errors: OmniContext: Show Status
4. Restart daemon if hit rate suddenly drops

Troubleshooting

Daemon Not Connecting

Symptoms:

• Sidebar shows "Daemon: Offline"
• Cache status shows "Offline"
• No context injection in AI chat

Solutions:

1. Check if daemon is running: OmniContext: Show Status
2. Manually start daemon: OmniContext: Start Daemon
3. Check daemon logs in Output panel (View → Output → OmniContext)
4. Verify binary path in settings: omnicontext.binaryPath
5. Restart VS Code

Common causes:

• Binary not found in PATH
• Port conflict (daemon already running)
• Insufficient permissions
• Antivirus blocking execution

Cache Not Working

Symptoms:

• Hit rate always 0%
• All requests show [fresh search] indicator
• Cache size shows 0/100

Solutions:

1. Verify pre-fetch is enabled: omnicontext.prefetch.enabled = true
2. Check daemon connection (see above)
3. Clear cache and restart: OmniContext: Stop Daemon → OmniContext: Start Daemon
4. Check for errors in Output panel
5. Verify workspace is indexed: OmniContext: Show Status

Common causes:

• Pre-fetch disabled in settings
• Daemon not connected
• Cache TTL too short (entries expire immediately)
• Workspace not indexed yet

High Memory Usage

Symptoms:

• VS Code using excessive memory (>500MB for extension)
• System slowdown
• Cache size growing unbounded

Solutions:

1. Reduce cache size: Set omnicontext.prefetch.cacheSize to 50-100
2. Reduce cache TTL: Set omnicontext.prefetch.cacheTtlSeconds to 180-300
3. Disable pre-fetch temporarily: Set omnicontext.prefetch.enabled = false
4. Restart daemon to clear memory: OmniContext: Stop Daemon → OmniContext: Start Daemon

Expected memory usage:

• Extension: 50-100MB
• Daemon: 100-200MB (depends on index size)
• Cache: ~1-5MB per 100 entries

Performance Issues

Symptoms:

• Slow context injection (>2 seconds)
• UI freezing during searches
• High CPU usage

Solutions:

1. Increase debounce delay: Set omnicontext.prefetch.debounceMs to 500-1000
2. Reduce cache size to decrease memory pressure
3. Check if indexing is in progress: OmniContext: Show Status
4. Verify system resources (CPU, memory, disk)
5. Profile with VS Code Developer Tools: Help → Toggle Developer Tools

Common causes:

• Large workspace (>10k files) still indexing
• Insufficient system resources
• Disk I/O bottleneck
• Too many concurrent searches

IPC Connection Errors

Symptoms:

• "Failed to connect to daemon" errors
• Reconnection attempts in logs
• Intermittent cache failures

Solutions:

1. Check named pipe/socket permissions (platform-specific)
2. Verify no firewall blocking local connections
3. Restart daemon: OmniContext: Stop Daemon → OmniContext: Start Daemon
4. Check for multiple daemon instances: Kill all and restart
5. Review IPC logs in Output panel

Platform-specific:

• Windows: Check named pipe \\.\pipe\omnicontext-ipc
• macOS/Linux: Check Unix socket /tmp/omnicontext-ipc.sock

Architecture

Event Flow

IDE Event (file open, cursor move, edit)
  |
  v
EventTracker (debouncing)
  |
  v
SymbolExtractor (get symbol at cursor)
  |
  v
IPC Client (send to daemon)
  |
  v
Daemon Pre-Fetch Cache (check cache)
  |
  v
  +-- Cache Hit --> Return cached context [cached]
  |
  +-- Cache Miss --> Search index --> Cache result --> Return context [fresh search]

Components

• EventTracker: Monitors VS Code events with debouncing
• SymbolExtractor: Extracts symbols at cursor using language features
• IPC Client: Communicates with daemon via named pipe/socket
• CacheStatsManager: Retrieves and formats cache statistics
• SidebarProvider: Renders sidebar UI with cache controls

Development

Building from Source

cd editors/vscode
npm install
npm run compile

Running in Development

1. Open editors/vscode in VS Code
2. Press F5 to launch Extension Development Host
3. Test features in the new window

Running Tests

npm test

Performance Metrics

Target performance (measured on typical developer machine):

• Event processing: <5ms
• IPC round-trip: <10ms
• Symbol extraction: <20ms
• Sidebar refresh: <100ms
• Memory overhead: <50MB

Contributing

See CONTRIBUTING.md for development guidelines.

License

Apache 2.0 - See LICENSE for details.

Support

• Issues: GitHub Issues
• Documentation: Main README
• Installation: INSTALL.md