@nomark-ai/pro
v0.1.1
Published
NOMARK Pro — trust contract, instinct engine, governance, critical-field gate
Maintainers
Readme
@nomark-ai/pro
Enterprise extensions for the NOMARK engine. Adds trust contracts, instinct capture, governance lifecycles, critical-field gates, and team sync on top of @nomark-ai/engine.
Install
npm install @nomark-ai/engine @nomark-ai/proRequires Node.js 18+ and @nomark-ai/engine >= 0.1.0.
Quick Start
import { createEngine } from '@nomark-ai/pro'
import { parseLedger } from '@nomark-ai/engine'
const entries = parseLedger(ledgerContent)
const engine = createEngine({
entries,
context: 'code',
trust: true,
instincts: true,
governance: true,
criticalFields: true,
})
// Process input through the full engine
const result = engine.process('draft the email to the client', {
schema: 'communication',
fields: ['recipient', 'tone', 'content_facts'],
})
// Check which fields need human confirmation
for (const gate of result.gate ?? []) {
if (gate.tier === 'critical_ask') {
console.log(`Confirm required: ${gate.field}`)
}
}Modules
Trust Contract
Governs agent autonomy based on demonstrated reliability. Trust score starts at 1.0, earned in drops (verified story completions) and lost in buckets (breach events S0–S4).
| Autonomy Level | Score Range | Restrictions | |----------------|-------------|-------------| | Full | 1.5+ | Unrestricted | | Trusted | 1.0–1.49 | Normal operation | | Supervised | 0.5–0.99 | Mandatory verification | | Restricted | 0.2–0.49 | No autonomous dispatch | | Probation | 0.0–0.19 | Owner confirmation required |
Instinct Engine
Pattern capture with confidence lifecycle. Instincts progress from pending through proven to promoted, with observation counting and confidence tracking.
import { createInstinctStore, captureInstinct, reinforceInstinct } from '@nomark-ai/pro'
let store = createInstinctStore()
store = captureInstinct(store, 'fmt-01', 'prefers bullet points', ['format'], now)
store = reinforceInstinct(store, 'fmt-01', now) // confidence increasesGovernance Lifecycle
Stage-based execution with trust gates, artifact tracking, and audit trails. Ships with a default CODE_LIFECYCLE config or accepts custom stage definitions.
import { createLifecycle, advanceStage, CODE_LIFECYCLE } from '@nomark-ai/pro'
const lifecycle = createLifecycle(CODE_LIFECYCLE, new Date().toISOString())
// Advance through stages: think → plan → build → verifyCritical-Field Gate
Schema-level inference tiers that prevent autonomous agent action on high-risk fields:
| Tier | Name | Agent Behavior | |------|------|---------------| | 1 | Inferable | Resolve from preferences | | 2 | Defaultable | Use defaults, allow override | | 3 | Critical Ask | Must confirm with human |
Built-in schemas for creative, decision, and communication request types. Extensible with custom schemas.
Team Sync & Admin
Multi-user team management with shared preference layers and Supabase-backed sync.
License
Business Source License 1.1
