Trigger.dev Release Notes

Records, per run, which worker cluster (cell) and lane (container or microvm) the most recent attempt executed in, so runs can be attributed to the cluster that served them and queried by it.

The supervisor adds both values to every attempt start it proxies for a runner, overriding anything the runner sends. The engine merges them into the run's existing annotations inside the update that already starts the attempt, replacing the previous stamp as a unit. Replication projects them into two new task_runs_v2 columns, worker_cell and worker_lane. The cell name comes from the new optional TRIGGER_WORKER_CELL_NAME supervisor env var (helm: supervisor.config.cellName); the lane is derived from the selected backend. Both also go into the worker instance metadata sent on connect, and admins see them on the run inspector. Neither is exposed to customers.

Mono-RevId: 87b2c6a5ce7724b4d32c2826fc74c266e3604c3a

1 breaking change, 5 improvements, and 6 server changes.

Highlights

Session creation now requires explicit trigger permissions

Creating or resuming a session with POST /api/v1/sessions now requires both session-write and task-trigger permissions. Scoped keys and public tokens must grant write:sessions (or access to the target session) and permission to trigger its task.

If you're using scoped API keys or public tokens to start or resume chat sessions, check that those keys include trigger permissions for the session's task.

Improvements

• chat.createStartSessionAction, chat.headStart, and chat.startHeadStart now accept a tags option for the Session itself, so chat sessions can be filtered by tag on the Sessions page. triggerConfig.tags still tags the session's runs.

await startChatSession({
  chatId,
  clientData,
  tags: [
    `org:${org.slug}`,
    `user:${user.id}`
  ],
});

• trigger.dev orgs list lists the organizations you belong to, with their slugs and IDs. Use the slug with trigger.dev projects create --org, including for organizations that don't have any projects yet.
• Reports filed with the submit_feedback MCP tool now reach the Trigger.dev team. They were being sent previously, but not somewhere anyone was reading.
• Stopping a chat.agent turn whose run() returns a streamText result no longer sends an error chunk with "An unexpected error occurred". The turn ends cleanly and the run stays alive for the next message.
• Update the minimatch dependency to ^10.2.3.

Server changes

These changes are included in the v4.7.3 Docker image and are already live on Trigger.dev Cloud:

• Self-hosted instances can block sign-in and sign-up from specific email domains with the new BLOCKED_EMAIL_DOMAINS environment variable.
• Run traces now appear almost immediately and load progressively as you watch, and much larger traces are viewable than before.
• The Logs page loads and searches faster, and busy environments no longer fail to load it. Searches that are too expensive now show a message suggesting a narrower time range instead of a generic error.
• Fix a rare case where a run could stay stuck in an executing state after the child run or wait it was waiting on had completed.
• Deployments with an unsupported runtime now fail with a 400 that names the supported runtimes, instead of a generic 500.
• Preserve the cancellation reason when a run that was executing is finalized, instead of replacing it with the default text.

How to upgrade

Update the trigger.dev/* packages to v4.7.3 using your package manager:

npx trigger.dev@latest update # npm
pnpm dlx trigger.dev@latest update # pnpm
yarn dlx trigger.dev@latest update # yarn
bunx trigger.dev@latest update # bun

Self-hosted users: update your Docker image to ghcr.io/triggerdotdev/trigger.dev:v4.7.3.

trigger.dev v4.7.3

Upgrade

npx trigger.dev@latest update # npm
pnpm dlx trigger.dev@latest update # pnpm
yarn dlx trigger.dev@latest update # yarn
bunx trigger.dev@latest update # bun

Self-hosted Docker image: ghcr.io/triggerdotdev/trigger.dev:v4.7.3

Release notes

Read the full release notes: https://trigger.dev/changelog/v4-7-3

What's changed

Breaking changes

Creating or resuming a session with POST /api/v1/sessions now requires both session-write and task-trigger permissions. Scoped keys and public tokens must grant write:sessions (or access to the target session) and permission to trigger its task.

Improvements

Update the minimatch dependency to ^10.2.3. (5236799b7)

Add trigger.dev orgs list to show the organizations you belong to, with their slugs and IDs. Use the slug with trigger.dev projects create --org, including for organizations that don't have any projects yet. trigger.dev projects create now activates the Free plan for organizations that don't have a plan yet, matching trigger.dev init. (68ee6be4b)

Reports filed with the submit_feedback MCP tool now reach the Trigger.dev team. They were being sent, but not somewhere anyone was reading. (4a94982a1)

Stopping a chat.agent turn whose run() returns a streamText result no longer sends an error chunk with "An unexpected error occurred". The turn ends quietly again and the run stays alive for the next message. (64c3ba13c)

chat.createStartSessionAction, chat.headStart and chat.startHeadStart now accept tags for the Session itself, so chat sessions can be filtered by tag on the Sessions page. triggerConfig.tags still tags the session's runs. (9dbc5f984)

await startChatSession({
chatId,
clientData,
tags: [`org:${org.slug}`, `user:${user.id}`],
});

Server changes

These changes affect the self-hosted Docker image and Trigger.dev Cloud:

Self-hosted instances can block sign-in and sign-up from specific email domains with the new BLOCKED_EMAIL_DOMAINS setting.

Run traces now appear almost immediately and load progressively as you watch, and much larger traces are viewable than before.

The Logs page loads and searches faster, and busy environments no longer fail to load it. Searches that are still too expensive now show a message suggesting a narrower range instead of a generic error.

Fix a rare case where a run could stay stuck executing after the child run or wait it was waiting on had completed.

Deployments with an unsupported runtime now fail with a 400 that names the supported runtimes, instead of a generic 500

Preserve the cancellation reason when a run that was executing is finalized, instead of replacing it with the default text.

All packages: v4.7.3

@trigger.dev/build, @trigger.dev/core, @trigger.dev/python, @trigger.dev/react-hooks, @trigger.dev/redis-worker, @trigger.dev/rsc, @trigger.dev/schema-to-json, @trigger.dev/sdk, trigger.dev

Contributors

triggerdotdev-the-inspector[bot], Eric Allam, Chris Arderne, Wes Mason, @nicktrn, @Mariefay, github-actions[bot], Oskar Otwinowski, James Ritchie, @isshaddad

Full changelog: v4.7.2...v4.7.3

Summary

1 new feature, 6 improvements, 4 bug fixes.

Breaking changes

Creating or resuming a session with POST /api/v1/sessions now requires both session-write and task-trigger permissions. Scoped keys and public tokens must grant write:sessions (or access to the target session) and permission to trigger its task.

Improvements

Update the minimatch dependency to ^10.2.3.
(5236799b7)

Add trigger.dev orgs list to show the organizations you belong to, with their slugs and IDs. Use the slug with trigger.dev projects create --org, including for organizations that don't have any projects yet.

trigger.dev projects create now activates the Free plan for organizations that don't have a plan yet, matching trigger.dev init.
(68ee6be4b)

Reports filed with the submit_feedback MCP tool now reach the Trigger.dev team. They were being sent, but not somewhere anyone was reading.
(4a94982a1)

Stopping a chat.agent turn whose run() returns a streamText result no longer sends an error chunk with "An unexpected error occurred". The turn ends quietly again and the run stays alive for the next message.
(64c3ba13c)

chat.createStartSessionAction, chat.headStart and chat.startHeadStart now accept tags for the Session itself, so chat sessions can be filtered by tag on the Sessions page.

triggerConfig.tags still tags the session's runs.
(9dbc5f984)

await startChatSession({
chatId,
clientData,
tags: [org:${org.slug}, user:${user.id}],
});

Server changes

These changes affect the self-hosted Docker image and Trigger.dev Cloud:

Self-hosted instances can block sign-in and sign-up from specific email domains with the new BLOCKED_EMAIL_DOMAINS setting.

Run traces now appear almost immediately and load progressively as you watch, and much larger traces are viewable than before.

The Logs page loads and searches faster, and busy environments no longer fail to load it. Searches that are still too expensive now show a message suggesting a narrower range instead of a generic error.

Fix a rare case where a run could stay stuck executing after the child run or wait it was waiting on had completed.

Deployments with an unsupported runtime now fail with a 400 that names the supported runtimes, instead of a generic 500

Preserve the cancellation reason when a run that was executing is finalized, instead of replacing it with the default text.

Raw changeset output

Releases

@trigger.dev/[email protected]

Patch Changes

Updated dependencies:
@trigger.dev/[email protected]
[email protected]

Patch Changes

Update the minimatch dependency to ^10.2.3.
(5236799b7)

Add trigger.dev orgs list to show the organizations you belong to, with their slugs and IDs. Use the slug with trigger.dev projects create --org, including for organizations that don't have any projects yet.

trigger.dev projects create now activates the Free plan for organizations that don't have a plan yet, matching trigger.dev init.
(68ee6be4b)

Reports filed with the submit_feedback MCP tool now reach the Trigger.dev team. They were being sent, but not somewhere anyone was reading.
(4a94982a1)

Updated dependencies:
@trigger.dev/[email protected]
@trigger.dev/[email protected]
@trigger.dev/[email protected]
@trigger.dev/[email protected]

Patch Changes

Updated dependencies:
@trigger.dev/[email protected]
@trigger.dev/[email protected]
@trigger.dev/[email protected]
@trigger.dev/[email protected]

Patch Changes

Updated dependencies:
@trigger.dev/[email protected]
@trigger.dev/[email protected]

Patch Changes

Updated dependencies:
@trigger.dev/[email protected]
@trigger.dev/[email protected]

Patch Changes

Updated dependencies:
@trigger.dev/[email protected]
@trigger.dev/[email protected]

Patch Changes

Updated dependencies:
@trigger.dev/[email protected]
@trigger.dev/[email protected]

Patch Changes

Stopping a chat.agent turn whose run() returns a streamText result no longer sends an error chunk with "An unexpected error occurred". The turn ends quietly again and the run stays alive for the next message.
(64c3ba13c)

chat.createStartSessionAction, chat.headStart and chat.startHeadStart now accept tags for the Session itself, so chat sessions can be filtered by tag on the Sessions page.

triggerConfig.tags still tags the session's runs.
(9dbc5f984)

await startChatSession({
chatId,
clientData,
tags: [org:${org.slug}, user:${user.id}],
});

Updated dependencies:
@trigger.dev/[email protected]
@trigger.dev/[email protected]

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

At Trigger we use AI as much as we can. We know you probably do too.
So today, we're adding a lot more AI into the Trigger dashboard - with the beta version of Ask Trigger - to help you debug, navigate, answer questions and even generate reports.
You can ask it stuff like:

• Why are my runs failing?
• Build a chart for all my failing runs

And it will also have page-specific context so you can ask "why is this run failing".
Hit Ask Trigger or ⌘J anywhere in the dashboard. Ask Trigger already knows where you are and can see your projects, environments, runs, errors, queues, tasks, and deploys across the organization.

Working through your data

The agent writes TRQL which it can use to query Trigger's analytics tables.
So it can answer any questions across all your runs and it isn't constrained by UI.
And since it has this data, the model can figure complex problems out for you - connecting different data tables.
It can also handle harder questions, like:

• Why is this run taking so long?
• Has this error happened before?
• How do I fix this error?

It will actually trace out the problem across the different data sources and show you the evidence to prove what it finds.
And if you connect your GitHub account it will inspect the source code too. And check the exact commit you deployed. Hone in on files and lines that broke and suggest a fix to you.

Generating reports

It can also generate completely custom reports for you.
For instance, asking "how's prod?" it will run a full health check across your project, checking all the different data and metrics sources, affected resources and any next steps that you can take.

Built with chat.agent

Built with chat.agent
Building this also gave us a chance to use chat.agent ourselves for a production agent.
chat.agent makes it easier to build complex chat agents that are durable and have features built in like response streaming. And they survive refreshes and resume if the connection drops mid-turn.
You can build your own agents with chat.agent.

It's open source

If you want to see how we built it, check out all the code here - it's open source.

Try it out

Ask Trigger is now in beta. Hit ⌘J from any project and ask it something real. Read more about the primitive behind it in the chat.agent docs.
We also know you want better tooling to give to your own AI agents. That's coming soon!

3 improvements and 1 server change.

Improvements

• Node.js deployments with runtime unset or set to "node" now use the current Node.js LTS. Set a versioned runtime like "node-24" to pin deployments to a specific Node.js major.
• chat.agent now sends parallel tool results to the model in the order the tools were called. Models that bind thinking to the exact conversation history (Claude Sonnet 5.5, Claude Opus 5.5, Claude Fable 5.1) no longer lose reasoning or fail when tools complete out of order.
• Recovering a long unfinished chat.agent response on a continuation run no longer blocks the worker. Response rebuilding is now linear in length and yields to the event loop, keeping heartbeats firing and preventing runs from being killed mid-replay.

Server changes

These changes are included in the v4.7.2 Docker image and are already live on Trigger.dev Cloud:

• Fix the "Download payload" and "Download output" buttons for large payloads and outputs, which previously downloaded an error file instead of the actual data.

How to upgrade

Update the trigger.dev/* packages to v4.7.2 using your package manager:

npx trigger.dev@latest update  # npm
pnpm dlx trigger.dev@latest update  # pnpm
yarn dlx trigger.dev@latest update  # yarn
bunx trigger.dev@latest update  # bun

Self-hosted users: update your Docker image to ghcr.io/triggerdotdev/trigger.dev:v4.7.2.

1 improvement and 1 server change.

Improvements

• trigger.dev init now stops with a clear error when the target directory has no package.json, instead of installing packages into a parent project. A custom task directory entered during interactive init is now written to trigger.config.ts relative to the project directory.

Server changes

These changes are included in the v4.7.1 Docker image and are already live on Trigger.dev Cloud:

• The CLI and MCP authorization page now shows the account and organizations you're granting access to.

How to upgrade

Update the trigger.dev/* packages to v4.7.1 using your package manager:

npx trigger.dev@latest update # npm
pnpm dlx trigger.dev@latest update # pnpm
yarn dlx trigger.dev@latest update # yarn
bunx trigger.dev@latest update # bun

Self-hosted users: update your Docker image to ghcr.io/triggerdotdev/trigger.dev:v4.7.1.

Upgrade

npx trigger.dev@latest update # npm
pnpm dlx trigger.dev@latest update # pnpm
yarn dlx trigger.dev@latest update # yarn
bunx trigger.dev@latest update # bun

Self-hosted Docker image: ghcr.io/triggerdotdev/trigger.dev:v4.7.2

Release notes

Read the full release notes: https://trigger.dev/changelog/v4-7-2

What's changed

Improvements

Use the current Node.js LTS for deployments whose runtime is unset or set to "node". Set a versioned runtime such as "node-24" to keep deployments pinned to that Node.js major. (9ecdf735e)

chat.agent now sends parallel tool results to the model in the order the tools were called, the same order the next turn's history uses. Models that bind their thinking to the exact conversation (Claude Sonnet 5.5, Claude Opus 5.5, Claude Fable 5.1) no longer lose a turn's reasoning, or fail the request on accounts that enforce the check, when its tools finish out of order. (760737258)

Recovering a long unfinished chat.agent response on a continuation run no longer blocks the worker. Rebuilding the response from its streamed chunks now takes time linear in its length and yields to the event loop as it goes, so heartbeats keep firing and the run is not killed mid-replay. Capturing the response at the end of a long turn gets the same speedup. (93c63c685)

Server changes

These changes affect the self-hosted Docker image and Trigger.dev Cloud:

Fix the "Download payload" and "Download output" buttons for large payloads and outputs, which downloaded an error file instead of the data.

All packages: v4.7.2

@trigger.dev/build, @trigger.dev/core, @trigger.dev/python, @trigger.dev/react-hooks, @trigger.dev/redis-worker, @trigger.dev/rsc, @trigger.dev/schema-to-json, @trigger.dev/sdk, trigger.dev

Contributors

Eric Allam, Chris Arderne, github-actions[bot], @Mariefay

Full changelog: v4.7.1...v4.7.2

trigger.dev v4.7.1

Upgrade

npx trigger.dev@latest update # npm
pnpm dlx trigger.dev@latest update # pnpm
yarn dlx trigger.dev@latest update # yarn
bunx trigger.dev@latest update # bun

Self-hosted Docker image: ghcr.io/triggerdotdev/trigger.dev:v4.7.1

Release notes

Read the full release notes: https://trigger.dev/changelog/v4-7-1

What's changed

Improvements

trigger.dev init now stops with a clear error when the target directory has no package.json, instead of installing packages into a parent project. A custom task directory entered during interactive init <path> is now written to trigger.config.ts relative to the project directory. (9499bd2a8)

Server changes

These changes affect the self-hosted Docker image and Trigger.dev Cloud:

The CLI and MCP authorization page now shows the account and organizations you're granting access to.

All packages: v4.7.1

@trigger.dev/build, @trigger.dev/core, @trigger.dev/python, @trigger.dev/react-hooks, @trigger.dev/redis-worker, @trigger.dev/rsc, @trigger.dev/schema-to-json, @trigger.dev/sdk, trigger.dev

Contributors

Chris Arderne, github-actions[bot]

Full changelog: v4.7.0...v4.7.1