Governance is the control module of the Govern pillar. It wraps around your existing Workbench agents with zero code changes and activates in Shadow Mode first, so you see exactly what it would block before anything is blocked. Agents built elsewhere join the same identity and policy through Gateway and AGDK.
Nothing about how your agents are coded changes. Governance runs beside the agent, not inside its code - which means adoption doesn't require a rebuild, and rollback doesn't require one either.
Shadow Mode evaluates every policy on every execution and logs what would have happened - without blocking anything. You see the blast radius before you ever flip the switch.
Emergency Stop blocks all agent executions across the entire workspace instantly, from the admin dashboard. No deployment required to pull the plug.
Policy definitions, gate decisions, HITL approvals, and identity records export in one click - built for the moment an auditor or regulator actually asks.
Matimo Governance is not a policy document - it's an execution-layer gate, an identity system, and an audit trail, wired directly into every Matimo Workbench agent.
Every agent action passes through: license check, Emergency Stop check, agent identity, lifecycle status, monthly governance budget, spawn-depth limit, and policy rules - evaluated in sequence, in real time.
License check Identity Budget Policy rulesAll policies evaluate on every execution. DENY decisions log as shadow_deny - nothing is actually blocked until you flip a single switch to Enforce.
Safe activationOne button in the admin dashboard blocks all agent executions instantly - in progress or new. Persists until an admin clears it. No deployment required.
Instant kill switchCryptographic ECDSA identity assigned to every agent - proves which agent performed which action, when. Revoking an agent takes effect within milliseconds.
ECDSA Non-repudiationA versioned, self-knowledge document per agent - identity, purpose, memory, and behavioral contract. Every change is SHA-256 hashed. Behavioral contract changes require HITL approval.
Versioned HITL-gated changes16 built-in policy templates, plus custom YAML rules tested against time, agent risk level, trigger source, LLM model, and spawn depth. Simulate against historical logs before publishing; conflict detection flags contradictions first.
16 templates SimulationA 0–100 score with an A–F grade, covering identity registration, Shadow vs Enforce status, policy coverage, Emergency Stop state, and shadow-deny rate. New tenants reach a healthy score in 15 minutes.
0–100 · A–F gradeSAML 2.0 single sign-on, verified end to end against Google Workspace, with JIT provisioning on first login. OIDC, LDAP sign-in and SCIM 2.0 provisioning are built, with live verification pending.
SAML 2.0 SSO SCIM 2.0PII and secret detection plus prompt-injection risk scoring on every request and response. Compliance Supporting Report for SOC2, HIPAA, GDPR, EU AI Act, ISO 27001 and NIST AI RMF. Evidence for your auditor, not a certification. Real-time event streaming to Splunk, Datadog, Elastic SIEM, or any webhook-capable tool.
PII detection 6 frameworks SIEM streamingGovernance sits between your Workbench agents and the outside world. It does not change how agents are built in Studio or via the API - it governs what happens when they run.
Every Matimo Workbench agent already inherits Matimo OSS's tool RBAC and audit logging. Matimo Governance adds the centralised policy registry, identity federation, and compliance evidence layer regulated enterprises need on top.
Explore Matimo Workbench →One agent identity, one policy history and one audit trail across every module.
One governed endpoint for every LLM call from any agent: identity, policy checks, routing and audit, with your own provider keys.
Content-safety screening for conversations and MCP tool calls: PII, secrets, prompt injection and unsafe content. Governance governs actions; Guardrails govern content.
An MIT-licensed Python SDK that gives agents built in LangChain, CrewAI, Google ADK, AutoGen or plain Python a cryptographic identity, telemetry and per-tool-call checks.
A 7-step governance gate, Shadow Mode safe activation, Emergency Stop, agent identity tokens, and one-click compliance evidence export across six frameworks - on agents that are already running.
SAML 2.0 SSO (verified against Google Workspace), JIT provisioning on first login, and OIDC, LDAP and SCIM support in the build - identity management your team already runs, extended to every agent.
Zero code changes to activate. Shadow Mode shows the blast radius before enforcement. Governance Health Score gives your team a single number to track as coverage matures.
Request Matimo Governance and activate Shadow Mode on your existing Matimo Workbench agents - no code changes, no risk of blocking anything until you're ready.
Already on Matimo Workbench? Add Matimo Governance to your workspace.