@plasius/ai-governance
v0.1.3
Published
AI guardrail, policy decision, confidence, and audit contracts for Plasius agentic AI.
Maintainers
Readme
@plasius/ai-governance
AI guardrail, policy decision, confidence, and audit contracts for Plasius agentic AI.
Scope
This package defines policy decision contracts for allow, deny, escalate, redact, and audit-only outcomes.
- decisions are resolved against confidence and data classification inputs
- feature-flag snapshots can force audit-only rollout behavior
- each outcome includes structured audit metadata for downstream compliance and replay
Install
npm install @plasius/ai-governanceUsage
import { packageDescriptor } from "@plasius/ai-governance";
console.log(packageDescriptor.packageName);import {
AI_GOVERNANCE_FEATURE_FLAGS,
resolveAiGovernanceDecision,
} from "@plasius/ai-governance";
const result = resolveAiGovernanceDecision({
requestedDecision: "allow",
policyId: "policy-default",
policyVersion: "2026-05",
correlationId: "corr-001",
confidence: 0.92,
dataClassification: "public",
featureFlags: {
[AI_GOVERNANCE_FEATURE_FLAGS.decisions]: true,
},
});
console.log(result.outcome);Development
npm install
npm run build
npm test
npm run test:coverage
npm run pack:checkRelease Workflow
Protected main releases use a two-step flow:
- Run
.github/workflows/cd.ymlwithbump=patch|minor|majorto open or refresh arelease/vX.Y.Zprep PR. - Merge that PR to
main. - Rerun
.github/workflows/cd.ymlonmainwithbump=noneto tag, draft the GitHub release, and publish to npm.
Governance
- Security policy: SECURITY.md
- Code of conduct: CODE_OF_CONDUCT.md
- ADRs: docs/adrs
- CLA and legal docs: legal
License
Apache-2.0
