Copy page

Release Notes

Release Notes — 2.71

Covers the 2.71 release series. Most recent patch: 2.71.12.

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

New features & improvements

SaaS & self-hosted

• Resource sharing — now generally available — share components and resources with the teams in your organization so they can use them in their own pipelines without duplicating configuration or handling secrets directly. Docs → (2.71.5)
• Kafka output — JSON-array bundling — the Kafka output can now bundle multiple records into a single JSON-array payload, cutting per-message overhead on high-volume streams. Docs → (2.71.1)
• Per-node retry queue viewer — you can now inspect a pipeline node's retry queue directly in the UI to see records awaiting retry and diagnose delivery issues. (2.71.3)
• Refreshed navigation — an updated navigation bar and user menu make it quicker to move around the app. (2.71.5)
• Searchable connector catalogs in the Monad MCP server — the MCP server's type catalogs are now searchable and bounded (no more oversized responses), and a new get_enrichment tool exposes enrichment details. (2.71.0) (For teams using the Monad MCP integration.)
• Version-aware docs links — in-product documentation links now open the docs for your deployment's version. (2.71.3)
• Google Cloud Storage input — schedulable polling — the Google Cloud Storage input now supports a polling schedule, so you can slow it down to stay within an upstream source's rate limits instead of polling continuously. Docs → (2.71.5)
  • Set the Cron Schedule field on the input's create/edit form; leave it blank to keep the default continuous polling.
• Cleaner errors in embedded connector forms — connector forms in embedded Monad now show validation problems inline on the offending field instead of a raw API response, for a more polished setup experience. (2.71.6) (For teams using Monad's embedded UI.)
• Expanded Monad MCP documentation — the docs now include a dedicated Model Context Protocol (MCP) section with setup guides for several MCP clients. Docs → (2.71.3) (For teams using the Monad MCP integration.)
• Richer schema-drift detection — schema-drift detection now reports per-field record counts and can optionally include parent objects, giving more context on what changed. (2.71.8)
• Clearer MCP tool metadata — every Monad MCP tool now advertises standard annotations (for example, read-only hints) so MCP clients can present the tools more accurately. (2.71.6) (For teams using the Monad MCP integration.)
• AWS SQS S3 inputs — FIPS endpoints for GovCloud — the SQS S3, SQS S3 CloudTrail, and SQS S3 GuardDuty inputs can now route STS, SQS, and S3 traffic to AWS FIPS 140-2 endpoints, so you can ingest from AWS GovCloud accounts (us-gov-east-1, us-gov-west-1) as well as commercial regions. Docs → (2.71.10)
  • Access: opt-in — turn on Use FIPS Endpoints in the input's settings. Supported in US, Canada, and GovCloud regions only.

SaaS only

• Trial period indicator — a chip in the navigation bar shows your trial status at a glance, with a link to manage your subscription. (2.71.7)

Self-hosted only

• Workload priority classes — platform and stateful components now run with dedicated Kubernetes priority classes, protecting the data path from resource contention under load on your cluster. (2.71.0)
• Self-managed metrics reference — new documentation describing the metrics a self-managed Monad deployment exposes, to help you monitor it. Docs → (2.71.8)

New connectors

• Braintrust — Audit Logs (input) — a new input that ingests audit logs from Braintrust, capturing account and workspace activity for your security data pipelines. Docs → (2.71.1)
• IBM QRadar (output) — a new output that delivers Monad-processed records to IBM QRadar (SIEM) for detection, search, and correlation. Docs → (2.71.8)
• Anthropic Claude Cowork (OTel) (input) — a new input that ingests OpenTelemetry telemetry from Anthropic's Claude Cowork, bringing its activity into your Monad pipelines for monitoring and security analytics. Docs → (2.71.10)
• AWS GuardDuty (SQS S3) (input) — a new input that ingests GuardDuty findings exported to S3 through the findings-export feature, picking them up via SQS notifications and emitting one finding per record, so your detections flow through the same pipelines as your CloudTrail data. Docs → (2.71.10)
• Coralogix (output) — a new output that forwards Monad-processed records to Coralogix for observability, search, and analytics. Pick your Coralogix region and, optionally, the record field to use as each event's timestamp. Docs → (2.71.10)

Fixes

• Component cloning keeps its version — duplicating a component now carries its version through, clearing a validation error some clone operations hit. (2.71.0)
• Create a component without pinning a version — component creation no longer requires an explicit version. (2.71.0)
• MongoDB inputs track progress per org and per project — the MongoDB org_events and proj_events inputs now keep a separate cursor per organization and per project, so events are no longer missed or duplicated across them. (2.71.1)
• SailPoint connection test authenticates — testing a SailPoint input connection now authenticates correctly, so the test result reflects real connectivity. (2.71.3)
• Pipeline node status updates immediately — a node's status now refreshes as soon as it changes instead of appearing stuck. (2.71.1)
• Transform boolean options show as toggles — boolean arguments in the transform builder now render as clear on/off toggles. (2.71.4)
• Monad MCP list responses report the true total — list results now show the real total count rather than the page size. (2.71.0)
• Monad MCP tool descriptions render correctly — repaired garbled characters (double-encoded UTF-8) in tool descriptions. (2.71.1)
• Syslog CEF/LEEF headers preserved — syslog inputs no longer strip the CEF/LEEF header out of the message, so ArcSight (CEF) and QRadar (LEEF) formatted events parse correctly. (2.71.7)
• Transform builder — Back and Cancel — the transform operation builder's Back and Cancel buttons now behave correctly instead of losing your place. (2.71.5)
• Land on the new item after duplicating — duplicating a resource now takes you straight to the new resource instead of leaving you on the original. (2.71.8)
• Azure Event Hub input no longer skips events — corrected cursor handling that could cause the Azure Event Hub input to skip records, so events are captured reliably. (2.71.10)
• CrowdStrike vulnerabilities input fixed — resolved an issue that could stop the CrowdStrike vulnerabilities input from pulling records. (2.71.9)
• Edge schema changes: removed fields no longer shown as additions — in a pipeline edge's schema-change view, a removed field is no longer mislabeled as a newly added field. (2.71.10)
• Parent-org API keys work for sub-org HTTP/OTel inputs again — an API key created in a parent organization can once more send data to HTTP and OpenTelemetry inputs whose pipelines live in one of its sub-organizations. (2.71.11)
• Permission changes reach sub-organizations immediately — changing a user's role or permissions in a parent organization now takes effect in its sub-organizations right away. (2.71.12)
• Org switcher remembers your last-active organization — after signing in, the app now opens the organization you were last working in instead of your oldest one. (2.71.11)
• Editing an API-created output no longer errors — opening the edit form for an output created through the API (for example an S3 output) with an unset setting no longer throws a runtime error. (2.71.11)
• Transform editor: live sampling excludes the transform's own node — when editing a transform, the live-data sampler no longer offers pipeline nodes running that same transform, which produced a spurious "key already exists" error. (2.71.10)
• Databricks output retries resend the full payload — retried uploads after throttling or transient errors now carry the complete data instead of an empty body. (2.71.10)
• KV Lookup enrichment recovers instead of restarting — if the enrichment's backing KV store is missing, the node now reports the error in the org log and retries rather than restarting repeatedly. (2.71.11)
• CrowdStrike input descriptions corrected — the Event Stream input now states that it carries Falcon Insight XDR detection and incident summaries plus platform audit events, and Device Inventory is attributed to the Falcon Hosts API rather than Spotlight. (2.71.11)

Breaking changes

Self-hosted only

• Standalone secret service removed from the Helm chart — the chart no longer deploys the secret-service workload or its MONAD_ENCRYPTION_KEY external secret; secrets are now resolved in-process by the API under your organization encryption keys. Action: a self-hosted deployment must be running a 2.71.0–2.71.9 release (which migrates existing secrets to the new store) before upgrading to 2.71.12 or later; do not jump directly from a release earlier than 2.71.0. Existing self-hosted customers have already been contacted and migrated. After upgrading, remove the secret: block from your values. (2.71.12)

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.