unstale

npm / inngest-cli

inngest-cli

What changed in inngest-cli 1, from its official release notes, upgrade guides and advisories. Sources checked 2026-11-02.

6 breaking1 deprecations6 new APIs and options4 behavior and defaults2 tooling7 fixes and other

See which of these apply to your code, ranked, in your agent:

npx @unstale/cli init

@unstale/cli 0.2.9 on npm

inngest-cli 126 changes

Breaking changes 6

  • Trailing slashes no longer bypass event request limits.

    1.49.0release notes

  • Custom Connect state managers must add this method, with the unchanged-connection check and deletion performed atomically.

    1.47.0release notes

  • Event names containing NUL characters are rejected as invalid.

    1.47.0release notes

  • Function throttles now apply to execution retries even when an earlier attempt ran within the last five minutes.

    1.47.0release notes

  • Each metadata write replaces the previous value for its kind on its span instead of merging; userland writes of the same kind from separate calls now replace rather than merge.

    1.47.0release notes

  • Step metadata is validated like checkpoint and API metadata; reserved inngest.* kinds outside the allowlist and invalid score values are dropped instead of stored.

    1.47.0release notes

Deprecations 1

  • Use include=output instead of includeOutput to request optional output; includeOutput remains supported.

    1.47.0release notes

New APIs and options 6

  • Adds a CLI command to replay a function's runs by time range and status.

    1.49.0release notes

  • Adds a CLI command to resume a paused function.

    1.49.0release notes

  • REST v2 run-list endpoints accept camelCase include selectors; API documentation recommends deferredFrom while existing deferred_from requests remain supported.

    1.47.0release notes

  • Adds a CLI command to pause a function through the v2 API.

    1.47.0release notes

  • The dev server and self-hosted server accept per-name score metadata kinds and show them in run details.

    1.47.0release notes

  • The dev server and self-hosted server accept per-code warning metadata kinds and show them in run details.

    1.47.0release notes

Behavior and default changes 4

  • Replay queries include skipped runs and provide more accurate paging.

    1.49.0release notes

  • Connect accepts larger execution requests and replies over internal gRPC, and no longer retries transport message-size violations. Deploy updated gateway and executor receivers together to support messages above 4 MiB.

    1.48.0release notes

  • Updated how step.invoke interacts with execution caps in cloud, preserving step invokes after an execution cap.

    1.48.0release notes

  • Run scheduling with rate limits retains retry protection for 30 minutes between attempts, allowing retries across ten-minute delivery gaps without consuming the same constraint again.

    1.47.0release notes

Tooling 2

  • Cloud users can copy an exact-run debugging prompt from failed run details and investigate through their AI tool using Inngest Cloud MCP.

    1.48.0release notes

  • The dev server explains when Cloud sandboxes require a paid plan and links to the upgrade page.

    1.47.0release notes

Fixes and other changes 7

  • After executor failure, Connect resumes the original request lease or buffered response and releases worker capacity when the replacement executor receives the result. Coordinate rollout and rollback around drained in-flight Connect work.

    1.48.0release notes

  • Renewed worker capacity reservations now expire when requests are abandoned without explicit cleanup, preventing Connect workers from incorrectly remaining at capacity.

    1.48.0release notes

  • The event API ingests valid events in a batch that also contains invalid events and returns an empty string in ids for each rejected event.

    1.47.0release notes

  • CEL presence checks now evaluate against concrete data, so an existing parent is not reported absent merely because a descendant is missing.

    1.47.0release notes

  • Extended-trace cap 429 responses now include Retry-After, so SDKs stop retrying immediately instead of stalling requests.

    1.47.0release notes

  • Scheduling retries no longer cancel an existing singleton run before rejecting the retry as a duplicate.

    1.47.0release notes

  • Restarting with -u no longer bumps function versions when a named app's ID is still derived from its URL and the register request is unchanged.

    1.47.0release notes