Skip to main content

Midcore Shell

Your AI workspace, native and signed.

One downloadable binary. The Maestro agent at slot 0. Robotics designer + policy gateway + fleet command. The research engine. The accountant. Personal knowledge base. All on top of a signed ASAR archive with hash-chained audit, integrity verification at every boot, and a local-first compute router that keeps your bill predictable.

Everything Midcore, in one window

slot 0

Maestro

The agent. Multi-modal chat with cross-package action bridge, slash commands, evidence-grounded provenance.

slot 1

Brain (Personal KB)

Your local-first knowledge base. DLP scrub, capsule export, EKB1 encryption, hash-chained audit.

slot 2

Robotics

τ₀-WM policy gateway + fleet command + designer + simulation + telemetry + safety + intake.

slot 3

Research

Scout/Survey/Pharmacovigilance tracks, Monte Carlo, MIT-style 7-stage wizard, manuscript engine.

slot 4

Studio

Notepads, capsules, chat sessions, the full midcore.ai web experience native.

slot 5

Automations

Triggers, workflows, MCP apps, email/bank/folder integrations.

slot 6

Robotics Designer

Build, edit, validate virtual robots with physics preview + τ₀-WM-ready templates.

slot 9

Account

Profile, subscription, balance, payments, usage, API keys, team, security, and the Compute routing dial.

slot 10

Vibe Coding

Bundled VS Code-class editor for code work without leaving the shell.

Built for trust

Hash-chained audit on every action

Local and cloud operations both record a tamper-evident audit node. Verifiable offline; survives shell restarts.

ASAR integrity at boot

Production builds verify a signed manifest of every shipped file before opening any window. Tamper → refuses to start.

Signed WASM runtimes

Every WASM blob the local-compute plane loads ships with an RSA-PSS sidecar signature verified against the same key.

Sandboxed workers

Local runners run in Node Worker threads with eval:false, bundle-only spawn paths, 1 GB resource caps, and zero auth-token visibility.

Tamper guards in production

Devtools blocked, debug flags detected, inspector listeners flagged. Every finding fires an audit-chain entry.

Code-signed installers

Windows, macOS, Linux installers signed with Azure Trusted Signing CSP. No SmartScreen warnings on the third install.

Pricing — transparent and metered, in CAD

Midcore Shell is the full-platform edition — one paid account unlocks every product surface bundled in your tier. The shell binary is free to download; cloud operations (LLM inference, τ₀-WM policy infer, fine-tunes) draw from your prepaid balance, but the local-first router prefers your hardware so most workloads cost nothing per call. Every account gets one 14-day trial seeded with $20 CAD of compute credit.

  • Starter $19 CAD/mo, 1 seat. AI-employee baseline (Foundation, Conscience, Mind, Council, Babel, Research, Robotics) + $25 CAD/mo compute included.
  • Pro $49 CAD/mo, 5 seats. Adds Close, Pivot, Genome, Loom, Tracks (finance + ops). $75 CAD/mo compute. Recommended.
  • Team $99 CAD/mo, 25 seats + SSO. Adds LedgerEye, Levy, Vault Banking. $150 CAD/mo compute.
  • Enterprise — contact sales. SCIM, on-prem / sovereign-cloud, dedicated CS, custom seat-based pricing.

See the full pricing page and the per-operation compute pricing catalog.

What every install enforces

  • Sign-in on first open— the shell does not run without authentication. 401 on every API call to anonymous traffic.
  • Subscription gate— cloud operations refuse 403 on inactive subscriptions before any provider is contacted.
  • Balance gate — cloud holds refuse 402 when balance is insufficient. Local fallback engages automatically when the operation supports it.
  • Pre-authorise cap— the shell prompts you the moment a cloud op would exceed your configured session cap.
  • Hard cap on fine-tunes— jobs estimated above $5,000 require split or support contact; no fat-finger draining of balance.
  • Signed receipts— every cloud charge produces a signed receipt recorded on the project audit ledger. Replayable offline.

Read more