MCP and CLI access now signs in via OAuth with approval of account, environment, permissions, and session duration, where read allows inspection and write allows changes; OAuth sessions cannot read event keys or signing keys. API keys for CI and unattended jobs are created by org admins with the same permissions and can be limited to one environment, with calls lacking the required permission returning 403 or insufficient_scope.
Inngest Changelog
npx @buildinternet/releases get inngest-changelogSandboxes are in public beta for Pro and higher accounts on the TypeScript SDK, with step.sandbox starting an isolated microVM from inside a function in under a second to run commands. Create, commands, and destroy are durable steps on the run trace, enabled via sandboxMiddleware() from inngest/experimental at inngest 4.21.1 or newer.
Pro and above accounts can now burst to 3x their account concurrency limit for spikes, with a 60-minute monthly budget that refills at an account-specific time and resets each month. Bursts are free, function-level concurrency overrides a burst, and admins get email notifications at 25%, 50%, 75%, and 100% of the budget.
A new AI Overview dashboard turns OpenTelemetry-derived gen_ai.* metadata into a live view of AI usage, estimated cost, and call latency, filterable by time range or function with charts and sortable tables for top and most-expensive runs, steps, and sessions. It appears under a new AI section in the dashboard sidebar and requires OpenTelemetry setup to capture AI calls made with OpenAI, Anthropic, Google Generative AI, or Vercel AI SDKs.
Sessions now propagate to child runs: any run triggered by another run via event, invoke, or defer inherits its parent's sessions, so setting sessions once on the root run groups the whole tree on the Sessions page. The Sessions page shows run counts, failures, recent activity, and the functions involved.
Sessions make it easier to investigate related work across your environment. When sending events, attach a stable session key and ID with meta.sessions, then use the Sessions page to jump from a domain concept like a conversation, support ticket, or import to the runs it…
Scoring lets you attach a named quality signal to a function run, a step, or an experiment variant. It shines for AI evals: run an LLM-as-a-judge over a result and write the verdict back as a score with inngest.score({ name, value }), or defer the judge entirely so a slow model…
Experiments let you compare versions of a step on live production traffic, judged on the execution data only Inngest can see. Call group.experiment("your-id", { variants, select }) with the implementations you want to compare and a selection strategy.
Deferred Functions are a new way to launch independent background work from inside an Inngest function. Call defer("some-id", { function, data }) and the parent run keeps executing — the deferred run is fully independent, with its own retries, concurrency, step state, and typed…
We've improved Insights and added new datasources, functionality and syntax support.
Realtime is now a first-class part of the TypeScript SDK. Channels, topics, publishing, and subscribing all have an improved DX, along with an updated useRealtime React hook.
You can now zoom into specific time ranges in traces for long-running functions with short steps using the timeline header's time brush. Drag the handles to narrow the view, move the selection to pan across the timeline, click outside to expand it, or use the reset button to…
The TypeScript SDK v4 is now available in beta! Major improvements include:
Inngest has released Agent Skills — pre-built capabilities for AI coding agents including Claude Code, Cursor, and Windsurf that offer current guidance on Inngest development.
Our official Helm chart is now published as an OCI artifact for easy installation and upgrades for self-hosting Inngest.
Inngest introduced Durable Endpoints, a feature that brings durability capabilities directly into API handlers — extending retries, recovery, and observability beyond workflows into ordinary endpoint code.
The documentation now includes quick actions enabling users to copy or view the current page in Markdown format. Users can also initiate new conversations with markdown URLs in Claude, ChatGPT, and other LLM providers. This markdown conversion minimizes the context required for…
Inngest has released a developer preview of Checkpointing, a new orchestration engine capability that enables near-zero inter-step latency while maintaining full workflow durability.
We are excited to announce the release of MCP Integration, bringing AI-assisted development workflows to the Inngest dev server. Connect Claude Code, Cursor, and other AI assistants directly to your local Inngest functions for seamless testing and debugging.



