Zones & Workspaces
Up to 10 governed terminal sessions in a configurable grid. 13 layout presets, drag-to-dock splitting, and per-zone health monitoring from pattern-matched output.
10 sessions · 13 presets · 4 monitorsThe first Managerial IDE
Direct AI agents. Govern every decision.
A local-first desktop environment where you direct AI agents through governed workflows instead of writing code by hand. Real terminal. Monaco editor. Multi-model AI. Everything on your machine.
- Local-first
- No telemetry
- Multi-model AI
- Beta opening soon
LYNS
⚔ cast review --staged
The Oracle studies your changes. Three findings, one warding concern in auth.ts.
The Council convenes ◆ convergent: 2/3
- 647 terminology keys
- 164 design tokens / theme
- 5,300+ passing tests
- 8 Relic content types
- 13 zone presets
- 0 cloud dependencies
Dual Identity
One engine. Two faces.
Flip the switch and the interface transforms. Every label, color, and voice remaps instantly. The same split lives inside the app. This very page just did it.
Quest Mode
Dev Mode
- The Oracle AI Chat
- The Council Multi-Agent Review
- Cast Run
- Quest Task
- Ward Audit
- Scrying Search
- Inscription Editor
- Relic Plugin Package
The whole site just remapped: theme, copy, and voice. The same switch lives in the app's header.
What's inside
Serious tooling, not a toy
A real engineering surface first. The campaign is optional. Both modes share the same engine, the same AI, and the same Relics.
AI Chat & Multi-Agent Review
The best model takes the Oracle seat and synthesizes. Cheaper or local models fill Council seats for independent analysis, with variance and convergence reporting.
Provider-agnostic · 1-3 Council seatsAutomated Audits
AI-powered codebase audits: tests, types, security, scope, rail receipts. Findings auto-reconcile on re-run and can be promoted to objectives.
6 built-in presetsPlugin Packages
Cryptographically signed content packages extend everything: skills, quests, themes, wards, formations, language servers. Every Relic is verified on install.
8 content types · Ed25519LSP Bridge
Real language intelligence: completions, diagnostics, hover, go-to-definition. Standalone language servers, crash recovery with backoff. Not a VS Code fork.
5 built-in serversLocal-first by default
Every session, profile, and Relic lives on your disk. No accounts, no telemetry, no cloud dependency. The system never guesses. It responds to explicit action.
0 cloud callsThe Relic ecosystem
Extend everything with Plugin Packages, signed and verified.
Relics are cryptographically signed content packages. Every one is verified with Ed25519 on install. Eight content types, one manifest, zero trust required.
- Skills
- Quests
- Themes
- Matchers
- Cast Config
- Wards
- Language Servers
- Formations
Marketplace One-click install from a signed catalog. The storefront opens in a later chapter. Creators keep their IP, buyers get verified packages.
Coming soon Quest Mode
Earned, never granted
The campaign layer is optional, and it never lies to you. XP comes only from real shell commands you actually type. Automated runs and Cast executions earn nothing. Progress you can trust.
Build Artificer
Forges systems and tooling. Lives in the build pipeline.
Stack Slayer
Full-stack campaigner. Slays bugs across every layer.
Test Warden
Guards the green suite. No regression passes the gate.
It begins with the Genesis Ritual: character creation on first launch. Then five onboarding quests teach the realm: casting skills, reading maps, raising wards, convening the Council, forging your own magic.
Sovereign by design
Your machine. Your rules.
Everything runs on your machine. Everything stays on your machine. No accounts to create, no telemetry to opt out of, no cloud to trust. Your keys, your models, your code, all yours.
- No accounts
- No telemetry
- No cloud sync
- Bring any model: Anthropic, OpenAI-compatible, or local Ollama
Beta access
Get early access
We're opening the beta soon. Drop your email and we'll notify you the moment it's ready.