Operator Tiers
Last updated: 2026-05-31
Five operator tiers. Each unlocks specific governance capabilities. The same TierFeatureMatrix that powers the SDK, the dashboard, and the CLI also powers this page. Queryable at GET /api/tiers (full matrix), GET /api/tiers/:tier (per-tier), POST /api/upgrade-hint (denial → remedy), GET /api/onboarding/:tier (per-persona quickstart).
Pick the lowest tier that covers what you need today. Upgrade when the next capability is required -- every denial response from the API includes the exact upgrade path. No surprises.
18,221 tests verifiedGovernanceBench 224/224VIGIL 148/148Kevlar OWASP 10/10
Tier 1 -- Evaluator
Evaluator
Auditor, regulator, first-time evaluator
TTFV target: 10 minutes to first benchmark report
Verifies claims, runs benchmarks, reads reports. Verifies certs and provenance chains. Runs the full benchmark battery against any deployment.
Identity
Verify a presented AgentCertificate
Trust
Query Sovereignty + Integrity Dyad for own agent
Authorization
Classify risk of a proposed action
Verification
Run GovernanceBench against your deployment · Run VIGIL adversarial battery · verify-cert / verify-provenance / verify-attestation CLI
Next: Upgrade to Analyst to unlock cert issuance for own agents + the full enforcement chain + Fleet / Audit / Compliance / Explainability / Trust Surface / Desktop / Chat dashboards + SIEM formatting + Sovereign Local Mode build. Quickstart:
docs/quickstart-evaluator.md Tier 2 -- Analyst
Analyst
Developer adding governance to an existing agent
TTFV target: 5 minutes to first governed action
Wraps an existing agent with @governed. Issues AgentCertificates for own agents. Sees the full read/observability dashboard surface. Formats events for SIEM. Builds entries in Sovereign Local Mode hash chains. Inherits every Evaluator feature.
Identity
Issue AgentCertificate for your own agent
Authorization
Full enforcement chain (classify + chain + compose) via POST /api/enforcement/evaluate
Audit
Format governance events for SIEM ingestion · Build entries in Sovereign Local Mode hash chain (SDK build-side) · MCP tool: audit-trail emission via agentomy_log
Secret
Format secret-fetch requests for 5 vault adapters (CyberArk CCP, HashiCorp Vault, Azure Key Vault, AWS Secrets Manager, GCP Secret Manager)
Verification
MCP tool: shadow AI exposure profile via agentomy_exposure_profile
Dashboard
Fleet · Audit Trail · Compliance · Explainability (EU AI Act Article 13) · Trust Surface (6 panels) · Desktop Daemon · Governed Chat
Next: Upgrade to Builder to unlock MCP attestation + provenance chain recording + cert issuance for delegated agents + workflow attestation + Slack integration. Quickstart:
docs/quickstart-analyst.md Tier 3 -- Builder
Builder
Engineer building agents at scale with custom policies
TTFV target: 30 minutes to first custom capsule policy + attested MCP
Issues certs for delegated agents. Attests MCP servers. Records cross-registry provenance. Posts to Slack via the governance gateway. Records workflow execution attestations. Inherits every Analyst feature.
Identity
Issue AgentCertificate for delegated agents
Attestation
Attest an MCP server manifest · Record cross-registry provenance entries · Hash-chained attestation of workflow execution
Audit
Workflow-scoped audit isolation key (cross-workflow leak prevention)
Operator
Send Slack message via governance gateway
Next: Upgrade to Operator to unlock halt + resume + quarantine + cert-bound operator sessions + fleet-wide dyad + live SIEM streaming + GDPR Article 22 view. Quickstart:
docs/quickstart-builder.md Tier 4 -- Operator
Operator
Ops / SRE running an agent fleet in production
TTFV target: 3 minutes from dashboard login to first halt-and-resume drill
Halts agents. Starts cert-bound operator sessions. Queries fleet-wide dyad + audit. Exports audit chains to regulators (hashes-only). Inherits every Builder feature.
Trust
Query Dyad across all governed agents
Authorization
Override risk classification per-action
Attestation
Start cert-bound operator session · Export-controlled session record (hashes-only)
Operator
Halt a single agent · Resume a halted agent · Quarantine an agent to scoped permissions
Audit
Read fleet-wide audit chain · Export audit chain to regulator · Live SIEM streaming
Dashboard
Steer Console (manual override) · GDPR Article 22 view
Next: Upgrade to Strategist to unlock fleet-wide halt + Dyad factor weight tuning + enforcement profile gate composition. Quickstart:
docs/quickstart-operator.md Tier 5 -- Strategist
Strategist
Executive, risk owner, board member, CISO
TTFV target: 5 minutes to executive summary
Fleet-wide halt authority. Tunes Dyad factor weights per vertical. Composes enforcement profile gates. Inherits every Operator feature.
Trust
Customize Dyad factor weights per vertical
Authorization
Tune enforcement profile gate composition
Operator
Halt the entire agent fleet (incident response)
