For teams using AxonFlow with Claude Code, Cursor, Codex, or OpenClaw who want longer audit trails, room for real policy packs, evidence export, and policy simulation without moving to Enterprise yet. Free 90-day license sent directly to your inbox. No sales call required.
Community is enough to install a plugin, write a few policies, and watch governance fire. Evaluation is the right next step once you're running a plugin against real work and Community's 3-day audit window and 20-policy ceiling start to bite.
Most teams reach this page after running a plugin in Community for a sprint or two and hitting one of the limits above. Evaluation is sized to validate plugin-driven governance against real work, not to spread across the org. Approval gates are not one of the things it unlocks: creating entries in the human approval queue needs Professional or higher.
Numbers below come directly from platform/agent/license/tier_support.go, the runtime configuration against which plugin requests are evaluated. Rows that don't apply to plugin users (MAP/WCP execution quotas, multi-agent plan caps, SSE streaming, cost-estimator limits) are omitted.
| Capability | Community | Evaluation (Free) | Enterprise |
|---|---|---|---|
| Tenant policies | 20 | 50 | Unlimited |
| Org-wide policies | 0 | 5 | Unlimited |
| Connectors with custom policies | 2 | 5 | Unlimited |
| Audit retention | 3 days | 14 days | Up to 10 years |
| HITL approval gates | — | Resolve-only — creating approvals needs Professional or higher | Unlimited, 24h expiry |
| Evidence export (CSV / JSON) | — | 5,000 records · 14-day window · 3 / day | Unlimited |
| Policy simulation | — | 300 / day | Unlimited |
| Session overrides (self-service unblock) | — | — | Enabled (24h cap, mandatory justification) |
Decision explainability (explainDecision) |
Available | Available | Available |
Per-user identity (X-User-Email) |
Available | Available | Available |
highRiskTools) pause for human approval before executing. Two independent things decide what happens. Whether the action is held follows the deployment mode: with DEPLOYMENT_MODE=community a require_approval match does not pause anything (the match is still recorded on the audit row), and every other mode holds it. Whether a reviewer entry is created follows the license tier and needs Professional or higher; below that the response reports approval_enqueue: "tier_disabled" and there is no reviewer entry for anyone to approve, so the action stays held and a retry is held again. Entries a deployment already holds stay resolvable from Evaluation up on the workflow and plan planes, so an upgrade never strands oneoverride_available: true on every block; this is what lets the user actually act on it. Enterprise-only.decision_id; explainDecision() returns the matched policies, risk level, and override availability for incident reviewX-User-Email, audit events and override grants are scoped to the actual human, not a synthetic shared client identityoverride_available: true on every block, but the create/use path is Enterprise-only), customer portal + SSO/SCIM (org-wide rollout to >5 developers), and up to 10-year audit retention (SOC 2 / ISO 27001 / HIPAA evidence windows).
See the full feature matrix for everything Enterprise unlocks.
Community is fully functional for installing the plugin and writing your first policies. Evaluation unlocks the capacity and the export / simulation features that matter once governance is firing on real work. Free, 90 days, renewable. AxonFlow runs fully self-hosted, and telemetry can be disabled for regulated evaluations.
decision_id, risk_level, and override_available on blocked actions)Most teams reach a real go / no-go decision within a few days of running the plugin against actual work.
If your organization needs to take one sponsored workflow into scoped production with founder-led rollout support, review the paid Production Program instead.
