Liminal documentation

Run, configure, and extend the Liminal agent harness (beta v0.1.2).

Published: docs.vireondynamics.com/liminal/

Recommended path

Install → Quickstart → Configuration basics
    → Connectors (if you use Gmail, Slack, Xero, etc.)
    → Architecture + Harness protocol
    → Persona + Vault guides (when you use those features)
    → Environment reference (lookup) + Troubleshooting (when stuck)

1. Get running

Step	Document
Install (CLI + desktop + auto-update)	Install
First session & smoke test	Quickstart
.env vs Settings vs defaults	Configuration basics
Something broke	Troubleshooting

Install commands (hosted):

curl -fsSL https://vireondynamics.com/install/install.sh | bash   # Linux / macOS / WSL

irm https://vireondynamics.com/install/install.ps1 | iex          # Windows

2. Understand the harness

Document	Contents
Architecture	Packages, ReAct loop, dispatcher, locks
Harness protocol	R-* rules, web_fetch, large writes
Runtime behavior	World context, drift, reflexion
Identity stack	Persona vs harness vs base LLM
UI streaming	SSE, busy state, web HUD
Rich message rendering	Markdown, HTML embeds, images in chat
Liminal desktop apps	spawn_app OS widgets (gated by AGENT_LIMINAL_APPS)

3. Integrations & work tools

Document	Contents
Connectors hub	Hosted OAuth, Integrations UI, agent tools
Multi-account mail	mail_search_inboxes, account_hint, redaction
Inbox watcher	Background Gmail/Outlook triage
Compose dock & attachments	Live file/email preview, drag-drop files
Google Workspace	Gmail, Calendar, Drive, Docs, Sheets
Microsoft 365	Outlook, Teams, Excel, OneDrive
Azure	ARM + Azure MCP
GitHub	Repos, issues, PRs
Slack	Channels, threads, post messages
Linear	Issues, projects
Notion	Pages, databases
Xero	Invoices, contacts, bills
YouTube	Channel, videos, analytics

4. Capabilities (how-to guides)

Document	Contents
Persona system · Persona bootstrap	Voice, soul slices, UI theme, first-run setup
Memory and vault · Memory & recall · Vault briefs	Typed notes, hybrid recall, scopes, curation, Obsidian
Web research	web_search + parallel web_fetch, retrieval order
Browser automation	Playwright: navigate, snapshot, act, extract
Document engine	PPTX / DOCX / PDF pipeline + quality gate
Voice	TTS + dictation
Sub-agents & orchestration	spawn_agent, critics, shared context
Dynamic workflows	Multi-phase parallel fan-out
Reasoning & effort	Reasoning budget vs output effort vs routing
Tuning via Settings	Change behavior without editing .env
Running the eval suite	Regression-test the harness
Sandbox capability lab	Isolated eval fixtures
Agentcard	Virtual cards, agent email, x402 payments
Remote sessions	Control a remote harness over the wire
Team memory · Team memory testing	Shared notes across Team tier

5. Pro & Enterprise

Document	Contents
Pro & Enterprise features	Per-tier features, entitlements, control-plane API
Accounts & licensing	Sign in, license storage, offline grace
Managed inference	Run with no API key (Pro)
Enterprise features	SSO, fleet, policy (operator guide)
Enterprise Edition	Open-core boundary, EE install + integrity
License	FSL-1.1-MIT (Community) + EE terms

6. Reference

Document	Contents
Changelog	Release notes through v0.1.2
Roadmap	Beta → RC → install GA
Environment reference	Generated full AGENT_* table (npm run docs:gen)
Configuration (narrative)	Flag groups by subsystem
Tool families	Full family catalog + lazy loading
Web API	REST + SSE, auth, multi-chat, integrations
Events	Telemetry payloads
Inference path validation	Diagnose routing / managed inference
Baseline profiles	Recommended flag bundles

Contributors

• Source: github.com/traidy2222/liminal-ai
• Agent-oriented map: root CLAUDE.md and README.md
• Local preview: npm run docs:dev in the monorepo
• After env key changes: npm run docs:gen · npm run docs:check
• Publish to production: in vireondynamics-website run npm run docs-portal:sync and deploy docs-portal