Changelog
Release history for governance-sdk. Follow on GitHub for updates.
v0.5.0latest2026-03-09
OpenAI Agents SDK adapter + enterprise module GA + 945+ tests
featOpenAI Agents SDK adapter — governAgent() for @openai/agents
featEnterprise module GA: RBAC, multi-tenant namespacing, fleet analytics
featPolicy templates: policyTemplates.fintech, .healthcare, .saas
featPolicy suggestion engine — agent-type-aware rule recommendations
featcreateEnterpriseGovernance() with org-level role management
featOrg analytics: fleetReport(), score trends, enforcement heatmaps
test945+ tests — 0 failures
docsFull quickstart guide at /docs with 8-step walkthrough
v0.4.02026-02-28
Event emitter, metrics, compose API — async governance lifecycle hooks
featGovernanceEmitter: on('enforce', 'register', 'kill', 'score') lifecycle hooks
featMetricsCollector: enforcement rates, block ratios, latency histograms
featcomposeGovernance() — composable governance instances with policy inheritance
featgov.on('blocked', handler) for real-time enforcement callbacks
perfenforce() average latency reduced from 0.8ms to 0.3ms (in-memory)
fixrequireSequence() now correctly handles concurrent tool calls
v0.3.02026-02-14
LangChain adapter + EU AI Act compliance module + PostgreSQL auto-migration
featLangChain adapter — governTool() wraps any DynamicTool with policy enforcement
featEU AI Act compliance module — assessCompliance(), getArticles(), getDaysUntilDeadline()
featPostgreSQL auto-migration — tables created on first connect, no migrations file needed
featcreateIntegrityChain() — HMAC-SHA256 hash-chained audit with verify() and export()
feataudit-integrity module: chain.append(), chain.verify(), chain.export()
test130 tests across 23 suites
v0.2.02026-01-31
Vercel AI SDK adapter + injection detection + kill switch
featVercel AI SDK adapter — createGovernedTools() wraps AI SDK tool definitions
featdetectInjection() — 64+ patterns across 7 categories (instruction override, exfil, role switch, obfuscation…)
featKill switch — kill(), killAll(), revive(), reviveAll() at priority 999
feat7-dimension governance scoring — composite 0–100 score, L0–L4 levels
featrateLimit(), tokenBudget(), timeWindow() policy conditions
featBoolean combinators — any_of(), all_of(), not() for complex rules
fixPolicy priority ordering now deterministic when priorities are equal
v0.1.0stable2026-01-15
Initial release — core policy engine, Mastra adapters, audit trail
featcreateGovernance() — core governance instance with rule evaluation
featblockTools(), allowOnlyTools(), requireLevel(), requireApproval(), requireSequence()
featMastra middleware adapter — createGovernanceMiddleware()
featMastra processor adapter — createGovernedMCPProcessor()
featgov.register() — agent registration with auto-scoring
featgov.enforce() — before-action policy evaluation with audit logging
featIn-memory storage — zero-config for development and testing
featTypeScript-native, zero runtime dependencies, MIT license
test81 tests across 14 suites at initial release
Stay up to date
Watch releases on GitHub Releases or subscribe to the npm feed.