Integrations

Works with your entire stack

20 official adapters across 4 categories. 2 storage backends. One import. Zero config. If you're building TypeScript AI agents, governance-sdk slots in.

20 adapters

4 categories

2 storage backends

0 config required

Framework Adapters

20 stable

AI Frameworks5

Model Providers4

Agent Platforms5

Cloud & Runtime6

Click any adapter above to see a code example

Storage Backends

2 stable

In-MemorycreateMemoryStorage()

(built-in)

Default. Zero config, zero setup. Perfect for development, testing, and CI.

10K event capacityZero configNo dependencies

// Default — no config needed
const gov = createGovernance({
  rules: [blockTools(['shell_exec'])],
});
// In-memory storage auto-created

PostgreSQLcreatePostgresStorage()

governance-sdk/storage-postgres

Production-grade persistence. Auto-creates tables on first connect.

Auto-migrateCustom table prefixSupabase / Neon / RDS

import { createPostgresStorage }
  from 'governance-sdk/storage-postgres';

const gov = createGovernance({
  storage: createPostgresStorage({
    connectionString: process.env.DATABASE_URL,
    tablePrefix: 'gov_',
  }),
  rules: [...],
});
// Tables auto-created. No migration needed.

Observability

built-in event system

Every enforcement decision emits structured events via the built-in GovernanceEmitter. Pipe these to any observability stack — Datadog, Prometheus, OTEL, Sentry, Grafana — with a few lines of code. No adapters needed.

gov.on("enforce", (event) => {
  // event.agentId, event.tool, event.decision,
  // event.latencyMs, event.rule, event.timestamp
  metrics.histogram("enforce_latency").observe(event.latencyMs);
});

gov.on("blocked", (event) => {
  alerting.send({ severity: "warn", ...event });
});

Missing your stack?

The adapter interface is public and takes ~50 lines to implement. Open an issue or PR.

Request adapter Read the docs