Roadmap
Current version: v0.5.2 — see the changelog for release history.
v0.6.0 — Scale & Operability
Remove scaling walls and make the agent more reactive.
- Informer-based ConfigMap watch replacing 3s polling in agent
- Downward API annotation reader — enables
stoker.io/ref-overrideand profile switching without pod restart - Per-gateway status ConfigMap sharding (eliminate write contention at 10+ gateways)
emptyDirsize limit on agent repo volume (prevent node disk pressure from large repos)- Webhook receiver rate limiting
v0.7.0 — Conditions & Validation
Operational visibility and safety for fleet management.
- New condition types:
AgentReady,RefSkew - Drift detection (re-sync same commit reports unexpected changes)
- Post-sync health verification (project state, tag providers — not just scan 200)
- Sync diff report in changes ConfigMap
- Conflict detection when multiple profiles map to the same destination path
- Validating admission webhook for GatewaySync CRs (reject invalid CRs at apply time)
- Structured audit logging (per-sync JSON record: timestamp, commit, author, gateway, files, result)
Future Ideas
These are valuable but not yet scoped into versioned milestones. They'll be prioritized based on user feedback.
Safety & Trust:
- Designer session project-level granularity (sync Project B while designer has Project A open)
- Pre-sync backup with auto-rollback on scan failure
- Module management (
.modlsync tomodules/withpostAction: restart) - Per-CR webhook HMAC secrets (replace global HMAC)
- Git commit signature verification (GPG/SSH, IEC 62443 compliance)
Reach:
- Standalone agent mode (systemd/Windows service for bare-metal Ignition servers)
- Approval annotation gate for production gateways
Enterprise:
- Maintenance windows and change freeze schedules
- External audit sink (SIEM integration via webhook/syslog)
- Drift detection with configurable action (report / restore / alert)
- Resource quotas and rate limiting for concurrent syncs