Skip to main content

Opta Status

Real-time visibility into every Opta service — what’s live, what’s built, and where to push it next.

Checking service status…
Inference Server
MLX-native inference on Apple Silicon
OpenAI-compatible /v1/chat/completions endpoint
Streaming SSE responses
GGUF model loading (llama.cpp fallback)
Automatic quantization selection
Model hot-swap without restart
Concurrent request handling
KV cache management
Context length enforcement
vLLM backend for parallel batching
Model Management
Model inventory API (/admin/models)
Dynamic load/unload API
Memory headroom enforcement (never crash on OOM)
Model health monitoring
HuggingFace model download integration
GGUF format support
Model benchmarking suite
API Compatibility
OpenAI /v1/chat/completions
OpenAI /v1/models
Health endpoint /healthz
Admin events SSE /admin/events
Rerank endpoint /v1/rerank
Skills API /v1/skills
Agents API /v1/agents
Embeddings endpoint /v1/embeddings
Performance
ANE (Apple Neural Engine) utilization
Batch request coalescing
Throughput metrics (tokens/sec)
Active request tracking
Voice & Audio
Speech-to-text endpoint POST /v1/audio/transcriptions (mlx-whisper)
Text-to-speech endpoint POST /v1/audio/speech (mlx-audio + Kokoro)
Multipart form upload for audio files (WebM, WAV, MP3)
soundfile I/O for audio processing
Share an idea for Opta LMX
Core Agent
Streaming agent loop with tool dispatch
Provider abstraction (LMX local + Anthropic fallback)
LMX connection with automatic discovery
Anthropic API integration (claude-sonnet-4-5)
Model routing by latency/capability
Context compaction at configurable threshold
Token counting + budget enforcement
TUI (Full-screen Terminal UI)
Full-screen Ink/React TUI (opta tui)
Chat mode with streaming output
Do mode (agentic, auto-approve safe tools)
Session switcher sidebar
Tool cards (collapsible, tool-call + tool-result)
Markdown rendering (code blocks, bold, lists, headers)
Slash command palette
Trigger mode detection (plan/review/research words)
TUI skill runtime with dynamic loading
Daemon
Background HTTP daemon (opta daemon start/stop/status)
WebSocket streaming endpoint (/v3/sessions/:id/stream)
Session persistence (JSONL event log)
Worker pool (parallel agent execution)
Bearer token authentication
Session management REST API
Background job runner (/v3/background/)
Operations registry (/v3/operations/)
Daemon metrics endpoint
Session replay API
LMX Integration
LMX client with auto-discovery
Model inventory and lifecycle management
/lmx scan — model catalog display
/lmx load / /lmx unload commands
/lmx status — health and active sessions
Model aliases (q, f, l → quality/fast/local)
Rerank API integration
Embedding pipeline
Browser Automation
Playwright MCP integration
Native session manager
Policy engine (approval gating)
Quality gates for regression detection
Visual diff manifests
Headless CI mode for browser tests
LSP Integration
LSP client (go-to-definition, references, hover)
Language server manager (start/stop per-project)
Protocol implementation (TypeScript, Python, Rust, Go)
Diagnostics in TUI
Code actions via LSP
Security & Config
Platform abstraction (macOS/Windows/Linux)
macOS Keychain via keyring (daemon token)
Windows DPAPI fallback keychain
Zod-validated config with self-healing
Per-tool permission model (allow/ask/deny)
Autonomy level enforcement (1-5)
Circuit breaker (warn/pause/hard-stop)
Policy runtime enforcement endpoint
MCP
MCP client (stdio + HTTP transports)
MCP server registry
Dynamic tool registration
MCP server health monitoring
Voice & Audio Operations
audio.transcribe daemon operation — routes base64 audio to LMX STT or OpenAI Whisper-1
audio.tts daemon operation — routes text to LMX TTS or OpenAI TTS-1
Provider selection from keychain (LMX local default, OpenAI cloud fallback)
Typed V3 protocol events: audio.transcription.result, audio.tts.chunk, voice.state
Input schemas: audioBase64, audioFormat, provider, language, voice
Cross-App Coordination
todo-optalocal/ cross-agent hub — structured handoff documents for multi-app feature changes
Share an idea for Opta CLI
Session Management
Daemon WebSocket streaming — real-time event consumption via WS
Session list — workspace rail with all active and past sessions
Session creation — start new sessions with model/mode selection
Session search — filter sessions by title or session ID
Session copy — duplicate session ID to clipboard
Session remove — delete sessions via daemon API
Token persistence — connection token stored in localStorage under opta:daemon-connection
Reconnect with backoff — exponential reconnect on WebSocket drop
Event cursor — resume from lastReceivedSeq to avoid duplicate events
Chat Interface
Markdown rendering — zero-dependency markdown renderer in MarkdownMessage.tsx
Code blocks — syntax-highlighted fenced code blocks
Inline formatting — bold, italic, inline code
Lists and headers — full Markdown block support
Blockquotes — styled quote rendering
Tool cards — collapsible tool-call and tool-result cards
Turn statistics — tokens, speed (tok/s), elapsed time, tool call count
Streaming indicator — live typing animation during assistant turns
Mode Toggle
Chat mode — standard conversational interaction
Do mode — agentic mode with auto-approval of safe tools
Mode indicator pill — header badge showing current mode
Per-session mode persistence — mode remembered across reconnects
Background Jobs
Background jobs page — view and manage long-running daemon processes
Job launcher — form to start new background processes with args
Job status — live status polling for active jobs
Job termination — kill running jobs
Connection Management
Connection settings — configure daemon URL and auth token
Connection status indicator — real-time online/offline badge
Auto-reconnect — transparent reconnection to daemon
Operations
Operations page — expose CLI-backed operations as GUI forms
Models page — view and manage LMX-available models
Settings page — daemon connection and app preferences
Packaging & Distribution
Electron/Tauri wrapper — native desktop app packaging
macOS DMG — distributable installer
Voice Dictation
useAudioRecorder hook — MediaRecorder API for in-browser audio capture
Mic toggle button in Composer — pulsating red ring while recording
Base64 WebM audio dispatch to daemon audio.transcribe operation
Transcription auto-appended to composer input draft
Spring physics animation on mic button (.r9-mic-recording CSS state)
Share an idea for Opta Code
Chat Interface
Streaming chat — token-by-token output via SSE
LMX model selection — choose loaded model from dropdown
Message history — persistent session conversation view
Stop generation — cancel in-flight inference
Code block rendering — syntax-highlighted code in responses
Dashboard
LMX connection status — production health is routed via /api/health; regressions are deployment-target related, not endpoint parity
Throughput metrics — tokens/sec via circular buffer (300 samples)
Active model display — currently loaded model name and size
Memory gauge — Metal GPU memory usage from /admin/health
Helper node status — embedding and reranking service health
SSE event stream — real-time dashboard updates from /admin/events
Session Management
Session list — browse and resume past sessions
Session creation — start new sessions with model selection
Session deletion — remove individual sessions
Authentication Modes
LAN mode — no auth required on local network
Cloud mode — Supabase auth via magic link / Google OAuth
Auth provider — useAuthSafe() returns null in LAN mode (no-op)
Sign-in page — redirect flow with sanitized next param
Session persistence — Supabase session cookie management
Settings
General settings — theme and display preferences
Tunnel configuration — Cloudflare Tunnel URL for remote access
Account settings — sign in / sign out, plan display
LMX endpoint override — custom LMX server URL
iOS App
iOS project foundation — SwiftUI app scaffold
Infrastructure
Vercel deployment — web app on optalocal.com
Cloudflare Tunnel — LAN services exposed for cloud access
Supabase auth — cloud-mode authentication backend
Share an idea for Opta Local
Landing Page
Hero section — value proposition and CTA
Download section — CLI installer links (macOS, versioned)
Stack overview — visual diagram of the Opta local stack
Feature highlights — key capabilities of CLI + LMX
Ecosystem links — connect to optalocal.com, status.optalocal.com
Setup Guides
CLI installation — step-by-step terminal instructions
LMX setup guide — install and configure LMX on Apple Silicon
Daemon setup — start background daemon and verify connection
First model download — guide to pulling the first LMX model
Interactive setup wizard — browser-based step-by-step configurator
OS detection — auto-detect macOS/Linux and show relevant guide
Documentation Pages
Architecture overview — how CLI, daemon, and LMX fit together
FAQ — common setup questions and troubleshooting
Changelog page — release history with dates
API reference — LMX and daemon API documentation
Technical
Static export — deployed to CDN via Vercel (output: 'export')
Sora + JetBrains Mono fonts — local font serving
Obsidian glass aesthetic — consistent Opta design system
Framer Motion animations — entrance animations and scroll effects
Responsive layout — mobile and desktop optimized
SEO meta tags — Open Graph and Twitter cards
Health badge — live status.optalocal.com indicator embedded in footer
RSS feed — subscribe to release announcements
Share an idea for Opta Init
Model Management
Live model inventory with load/unload controls
Memory usage per model with headroom visualisation
Model hot-swap (no restart required)
HuggingFace download integration
GGUF format display and management
Model alias configuration (q, f, l shortcuts)
Chat Interface
Browser-native chat with streaming output
System prompt editor per conversation
Conversation history and session switching
Markdown rendering in responses
Token usage display per request
Monitoring & Metrics
Real-time throughput gauge (tokens/sec)
Active request tracker
KV-cache pressure indicator
ANE (Apple Neural Engine) utilisation graph
Latency histogram per model
Sessions
Active session list with prompt previews
Session replay viewer (full event log)
Session kill / force-evict controls
Arena
Side-by-side model comparison interface
Shared prompt entry with split response panes
Per-model latency and token count comparison
Benchmarks
Benchmark suite runner (PPL, throughput, latency)
Historical benchmark chart (per-model trend)
Export benchmark results as JSON
Agent Monitoring
Running agent overview (source, model, status)
Per-agent event stream viewer
Agent kill controls
Bridge
LMX bridge connectivity status
Remote LMX endpoint configuration
Proxy mode toggle (route external requests to local LMX)
Diagnostics
System health panel (CPU, RAM, ANE, disk)
LMX process log viewer (tail last 200 lines)
Error registry with timestamps and stack previews
Skills
Installed skill list with version display
Skill enable/disable toggles
Skill detail viewer (description, parameters)
Settings
LMX endpoint URL configuration
Default model selection
Auto-refresh interval control
Theme switcher (HD / Lite)
Share an idea for Opta LMX Dashboard
Authentication & Identity
Next.js App Router — foundation with Supabase integration
Magic link authentication — passwordless email sign-in
OAuth providers — Google sign-in integration
Session management — Supabase cookie-based sessions
Profile management — user settings and account details
Ecosystem Integration
SSO portal — centralized login for optalocal.com and other apps
Capability model — role-based access to Opta features
Supabase backend — robust user data and schema management
Aesthetic & UI
Opta design system — consistent typography (Sora) and colors
Glassmorphism UI — elevated aesthetic and modern look
Responsive design — mobile-friendly authentication flows
Share an idea for Opta Accounts
Health Monitoring
Service status overview — live operational state of apps
LMX reference instance monitoring — configured via OPTA_LMX_HEALTH_URL (legacy fallback: OPTA_LMX_TUNNEL_URL) and reported as online/degraded/offline
Daemon reference instance monitoring — configured via OPTA_DAEMON_TUNNEL_URL / OPTA_DAEMON_STATUS_URL, with 401/403 deep-status probes treated as online + auth-required rather than false degraded/offline
Auth-boundary UI signaling — daemon cards surface Auth required for deep health when status is reachable but tokenless
Management websites tracking — Opta Local (optalocal.com), Init, Accounts, Help, Learn, and Admin
Priority policy: when management websites drift, those are treated as highest-priority repair tasks before single-surface feature work
Auto-refresh — polling every 30 seconds for live updates
Feature Registry
Per-app feature lists — parsed markdown checklists showing what's live
Service identity colors — each app has a distinct semantic accent
Idea / feature request form — inline per-app feedback submission
Release Notes
Automated generation — compiled from CLI and LMX update logs
Markdown parsing — frontmatter extraction and bullet summarization
Categorization — model runtime vs stack sync updates
Aesthetic & UI
Opta Ring component — animated CSS ambient singularity effect
Dark mode first — void background with neon indicators
Responsive layout — optimized for desktop and mobile
Share an idea for Opta Status
Purpose
Private internal workspace for website operations
Fleet visibility across management websites (Home, Init, Accounts, Status, Help, Learn, Admin)
Surface Opta Learn guide readiness + promotion controls
Promotion action audit visibility (structured recent events)
Status + feature-registry snapshot integration in admin cockpit
Evidence / Integrations
Dashboard shell (1A13-Opta-Admin)
Website fleet health probes (local + production parity)
Learn guide draft -> verified promotion endpoint
Admin ops snapshot API (/api/admin-ops) with recent action records
Status + feature-registry integration (status.optalocal.com/api/health/admin + docs/feature-audit.md ingest)
/api/health endpoint reachable in production for status polling
Share an idea for Opta Admin
Knowledge Base
Documentation structure — organized by application (CLI, Code, LMX)
Search functionality — quick access to specific guides and troubleshooting
Support & Guidance
Getting started guides — step-by-step setup tutorials
API references — technical details for developers
FAQs — answers to common user questions
Ecosystem & UI
Opta design system — aligned with the overall Opta aesthetic
Interlinked navigation — easy switching between Opta management websites
Responsive layout — accessible on all devices
Share an idea for Opta Help
Guided Onboarding
Structured guide taxonomy (CLI, LMX, Accounts, Code, Local Web)
Interactive getting-started learning paths
Cross-app glossary for shared architecture concepts
Progressive disclosure for new users (beginner → advanced)
Guide Engine
Typed guide registry (content/guides/index.ts)
Template families (feature deep-dive, masterclass, operation, configuration)
Guide generation workflow with prompt-driven inputs
Guide validation pipeline (content/guides/index.ts)
Search index integration (lib/search.ts)
Delivery
Fast static generation for guide content
Mobile-responsive reading and guide index UI
Keyboard-friendly navigation
Platform Integrations
Integration pointers to Opta CLI, LMX, and Code Desktop
Deep links into setup and operations workflows
Help/contact CTA to Opta support surfaces
Share an idea for Opta Learn
Opta Local ecosystem · auto-refresh 30sFeature registry