Queue Monitoring, Object Storage & Security Hardening

Here's what shipped in April. This one is mostly operational: better visibility into the queue, a sturdier storage layer, and a round of security fixes.

Queue monitoring

Instance admins can watch the queue from the admin dashboard. See depth and health at a glance, so you catch a backlog before it turns into late deliveries.

Object storage refactor

We reworked the object storage layer. Exports and large payloads now go through a cleaner path that's easier to reason about and less likely to break under load.

API versioning

We upgraded to request-migrations v2. Older API clients keep working as we evolve the API, so an upgrade on our side doesn't break your integration.

Security and reliability

• No token replay: password reset tokens are now single-use
• AMQP credentials: special characters in RabbitMQ connection credentials are escaped correctly
• SSO: hardened the login and retry-deactivation paths
• Split proxies: the webhook dispatcher honors NO_PROXY
• CI: nightly CVE scans flag vulnerable dependencies early

Update when you get a chance, then keep an eye on your queue from the admin dashboard.