Internal development guide for the NEXUS AI Workstation Builder.
# Clone
git clone https://github.com/AiImpDevelopment/nexus-workstation.git
cd nexus-workstation/Nexus
# Install dependencies
pnpm install
# Development mode (hot reload)
cargo tauri dev
# Run Rust tests
cd src-tauri && cargo test
# Build release
cargo tauri build --release23 modules organized by domain:
| Module | Purpose | Key Commands |
|---|---|---|
router |
Intelligent model routing (Ollama/OpenRouter) | route_message, route_message_stream |
orchestrator |
Standalone AI brain (MAPE-K, Hebbian trust) | Background tasks, scheduling |
chat |
Tauri Channel-based streaming | chat_stream |
docker |
Docker container management | list_containers, container_action |
github |
GitHub API integration | get_repos, get_issues, get_pull_requests |
agents |
Custom AI agent CRUD | create_agent, list_agents |
inference |
HuggingFace Hub + GGUF loading | cmd_download_model, cmd_load_gguf |
ide |
Filesystem + search + command execution | ide_read_file, ide_search_files |
browser |
Internal browser management | open_internal_browser |
cdp_engine |
Chrome DevTools Protocol (chromiumoxide) | cdp_navigate, cdp_screenshot |
cdp_network |
HTTP waterfall capture | cdp_network_entries |
cdp_devtools |
Console, performance, cookies | cdp_console_entries, cdp_perf_metrics |
browser_agent |
AI-powered web automation | browser_agent_run |
browser_import |
Bookmarks/history import | browser_detect_profiles |
web_scraper |
Built-in HTML scraper | web_scrape, web_extract_metadata |
theme_engine |
ElvUI-inspired UI customization | theme_save, theme_validate_contrast |
widget_registry |
Modular dashboard components | widget_list, widget_categories |
style_engine |
BenikUI sub-component styling | style_get_widget, style_save_graph |
evaluation |
Agent-as-a-Judge quality scoring | eval_agent_output |
settings |
App configuration persistence | cmd_get_settings, cmd_set_setting |
monitoring_quick |
Lightweight sysfs-based system stats | cmd_get_quick_stats |
system_agent |
Offline health checks | system_scan |
neuralswarm |
NeuralSwarm status bridge | neuralswarm_status |
error |
Structured error handling (NexusError) | 7 categories, panic hook |
| Route | Page |
|---|---|
/ |
Dashboard (redirect to chat) |
/chat |
AI Chat with model routing |
/docker |
Docker container manager |
/github |
GitHub repos, issues, PRs |
/n8n |
Automation workflows |
/news |
Tech news aggregator |
| Store | Purpose |
|---|---|
chat.svelte.ts |
Chat state, message history, streaming |
errors.svelte.ts |
Centralized error handling + toast queue |
# All Rust tests
cd src-tauri && cargo test
# Specific module
cargo test --lib browser_agent
cargo test --lib web_scraper
cargo test --lib theme_engine
# Current count: 144 tests- Rust:
cargo fmt+cargo clippy - TypeScript/Svelte: Prettier + ESLint
- Commits: Conventional Commits (
feat:,fix:,chore:) - Branching:
main(releases),develop(active dev)
- Standalone: No external service dependencies (PostgreSQL, Redis, systemd)
- Offline-first: Local Ollama models preferred, cloud as fallback
- MIT/Apache-2.0 only: All dependencies must be commercially safe
- Cross-platform: Linux (primary), Windows, macOS
- SQLite bundled: WAL mode, rusqlite, no external DB
- Structured errors: NexusError with categories, suggestions, and panic recovery
- Update version in
Cargo.tomlandtauri.conf.json - Run full test suite:
cargo test - Build:
cargo tauri build --release - Tag:
git tag v0.5.1 - Push:
git push origin main --tags - GitHub Actions builds cross-platform binaries
MIT — see LICENSE