Release Notes — 2.63

Covers the 2.63 release series. Most recent patch: 2.63.3.

Unless a subsection notes otherwise, changes apply to both Monad SaaS and self-hosted.

New features & improvements

• Alert rule status wording is now Enable / Disable — the alert-rule status control uses consistent Enable/Disable terminology across the app. (2.63.1)
• Manage alert rules directly from the list — each alert rule now has a per-row enable/disable toggle, shows its alert category, and marks whether it is system-managed, so you can see and change rule state without opening each one. (2.63.1)
  • Access: on by default — visible on the Alerting rules list.
• Hosted MCP service & Connected Apps — Monad now runs a hosted Model Context Protocol (MCP) service and adds a Connected Applications area with an OAuth org picker, so you can connect Monad to MCP-compatible AI clients and manage those connections in one place. (2.63.1)
  • Access: opt-in — connect and manage apps under Settings → Connected Applications.
• Custom pipeline edge names — You can give pipeline edges your own labels in the V2 pipeline builder, so branching flows read clearly at a glance. (2.63.1)
  • Access: opt-in — edit an edge's label in the pipeline builder.
• Firehose input: templating & connection pooling — The Firehose input adds a templating option plus connection-pool controls, giving you more flexible routing and better throughput on high-volume streams. Docs → (2.63.1)
  • Access: opt-in — configure templating and pool options in the Firehose input settings.

New connectors

• Method Financial — Accounts (input) — pull Method Financial account records into Monad pipelines, so account data can be normalized and routed alongside the rest of your security and audit telemetry. Docs → (2.63.3)

Fixes

• The API now returns 404 (not 500) when a resource doesn't exist — requesting a pipeline, input, transform, or output that was never created or has been deleted now returns a proper 404 Not Found instead of a generic 500. Tooling that keys off the status code — including Terraform — can now tell "gone" apart from "broken", and the Monad UI handles the new response. (2.63.3)
• Creating an alert rule with no pipelines selected no longer freezes the page. (2.63.3)
• Expand-all in the schema field tree — the expand-all control in the schema field tree now works correctly. (2.63.1)
• Org selection survives multiple sign-in replicas — Signing in no longer loses your selected organization when the login service runs more than one replica. (2.63.1)
• Azure Blob Storage input accepts arbitrary time formats — The Azure Blob Storage input now parses non-standard timestamp formats instead of failing on them. (2.63.1)
• Azure Event Hubs input state race fixed — A race condition when saving Event Hubs consumer state is resolved, preventing occasional state corruption. (2.63.1)

Breaking changes

Self-hosted only

• Outbound SSRF protection is now on by default — Monad's HTTP client now blocks outbound requests to private / internal network addresses (loopback, RFC1918 IP ranges, and internal hostnames) by default. Previously there was no such restriction, so a self-hosted deployment that routes a pipeline to an internal or private host — a private output, input, or enrichment endpoint on your own network — will find those connections blocked after upgrading. (Monad's own in-cluster calls, such as SSO to an in-cluster identity provider, are exempted automatically and are unaffected.) (2.63.0)
  • Action: before upgrading to 2.63.x, list the internal hosts you need to reach in the Helm value MONAD_ENV_ALL__MONAD_INTERNAL_HOSTS (forwarded into the pipeline pods) — and MONAD_INTERNAL_HOSTS on the api — as comma-separated hostnames, IPs, or CIDRs. Use * to allow all and restore the previous behavior.

Need help?

See the Monad docs for setup guides and reference, or reach out to Monad Customer Support at support@monad.com or via your dedicated Slack customer channel.