Liminal 0.0.16 — managed inference, dynamic routing & Pro sign-in

v0.0.16 is the current tip of liminal-ai main: Pro users can run with no provider API key, OpenRouter routing is price-sorted with sticky sessions and 429 rotation, and Enterprise Edition installs itself after sign-in. Community Edition stays local, bring-your-own-key, and free.

Shipped in this slice#

• Managed inference (Pro) — Chat, embeddings, vision, and audio sidecars can route through Vireon's metered OpenAI-compatible proxy (AGENT_INFERENCE_MODE: auto · managed · byok). Included monthly credits, optional top-ups, and a clear prompt when budget is exhausted. After liminal login, managed routing is preferred; switch back to BYOK in Settings anytime
• Vireon sign-in — liminal login (loopback browser) or Settings → Sign in stores your license under ~/.liminal/ — no .env key paste for Pro managed mode
• Dynamic OpenRouter routing — Price-sorted reseller selection (AGENT_PROVIDER_STRATEGY), sticky session_id for KV-cache affinity (AGENT_OPENROUTER_SESSIONS), and session-epoch rotation on upstream 429s (adaptive) so rate limits do not stall long ReAct runs
• Enterprise Edition auto-install — Pro+ login downloads the proprietary EE bundle to ~/.liminal/enterprise/ (checksum-verified); paid implementations stay out of the public CE repo
• Security hardening — Control-plane rate limiting and auth hygiene, web API token gate bound to loopback, tighter guards on run_shell and execute_code
• Inference UX — Web inference usage/credits banner; steadier sign-in polling for managed sessions
• Prompt cache + routing — AGENT_PROMPT_CACHE breakpoints pair with sticky OpenRouter sessions for cheaper multi-round turns when the provider supports cache reads
• Dictation & ASR defaults — Default AGENT_TRANSCRIBE_MODEL is nvidia/parakeet-tdt-0.6b-v3; AGENT_DICTATION_WEB_SPEECH=0 keeps browser Web Speech off so dictation transcribes server-side only (no Chrome→Google path)
• Technical release notes — Full operator bullets on the docs changelog

What you can run today#

npm run build -w packages/core && npm run build -w packages/tools
npm run web   # or: npm run tui

Hosted installer:

curl -fsSL https://www.vireondynamics.com/install/install.sh | bash

Sign in for managed inference (Pro):

liminal login

Docs: docs.vireondynamics.com/liminal/

Roadmap (not shipped)#

Pre-release history#