## Summary
`chat.agent` now takes a `tools` option. Until now tools only went to
`streamText` inside `run()`, so the SDK had no tools when it
re-converted the persisted `UIMessage` history at the start of each
turn. Any tool with a `toModelOutput` (raw image bytes into an image
content part, or a sub-agent transcript compressed to a summary) had its
transform applied on turn 1 and skipped from turn 2 onward, so the raw
output got JSON-stringified back into the prompt and the model lost the
transformed view.
Declaring `tools` on the config threads them into that conversion, so
`toModelOutput` runs on every turn. The resolved set is handed back,
typed, on the `run()` payload as `tools`:
```ts
const tools = { searchDocs, renderChart };
export const myChat = chat.agent({
tools,
run: async ({ messages, tools, signal }) =>
streamText({ ...chat.toStreamTextOptions({ tools }), messages, abortSignal: signal }),
});
```
`tools` also accepts a per-turn function for tools that depend on the
user or a feature flag. Only `inputSchema` and `toModelOutput` are read
during conversion, never `execute`. Also exports
`InferChatUIMessageFromTools<typeof tools>` to derive the chat
`UIMessage` type from a tool set. No behavior change for agents that
don't declare `tools`.
Trigger.dev References
Contains code that tests or uses the @trigger.dev/* packages in some way, either by using them to test out a framework adapter, an integration, or parts of the main SDK.
All the dependencies to the @trigger.dev/* packages will be both referenced in the package.json dependencies as workspace:*, as well as using a direct path from the tsconfig.json file like so:
{
"extends": "@trigger.dev/tsconfig/node18.json",
"include": ["./src/**/*.ts"],
"compilerOptions": {
"baseUrl": ".",
"lib": ["DOM", "DOM.Iterable"],
"paths": {
"@/*": ["./src/*"],
"@trigger.dev/sdk": ["../../packages/trigger-sdk/src/index"],
"@trigger.dev/sdk/*": ["../../packages/trigger-sdk/src/*"],
"@trigger.dev/express": ["../../packages/express/src/index"],
"@trigger.dev/express/*": ["../../packages/express/src/*"],
"@trigger.dev/core": ["../../packages/core/src/index"],
"@trigger.dev/core/*": ["../../packages/core/src/*"],
"@trigger.dev/integration-kit": ["../../packages/integration-kit/src/index"],
"@trigger.dev/integration-kit/*": ["../../packages/integration-kit/src/*"],
"@trigger.dev/github": ["../../integrations/github/src/index"],
"@trigger.dev/github/*": ["../../integrations/github/src/*"],
"@trigger.dev/slack": ["../../integrations/slack/src/index"],
"@trigger.dev/slack/*": ["../../integrations/slack/src/*"],
"@trigger.dev/openai": ["../../integrations/openai/src/index"],
"@trigger.dev/openai/*": ["../../integrations/openai/src/*"],
"@trigger.dev/resend": ["../../integrations/resend/src/index"],
"@trigger.dev/resend/*": ["../../integrations/resend/src/*"],
"@trigger.dev/typeform": ["../../integrations/typeform/src/index"],
"@trigger.dev/typeform/*": ["../../integrations/typeform/src/*"],
"@trigger.dev/plain": ["../../integrations/plain/src/index"],
"@trigger.dev/plain/*": ["../../integrations/plain/src/*"],
"@trigger.dev/supabase": ["../../integrations/supabase/src/index"],
"@trigger.dev/supabase/*": ["../../integrations/supabase/src/*"],
"@trigger.dev/stripe": ["../../integrations/stripe/src/index"],
"@trigger.dev/stripe/*": ["../../integrations/stripe/src/*"],
"@trigger.dev/sendgrid": ["../../integrations/sendgrid/src/index"],
"@trigger.dev/sendgrid/*": ["../../integrations/sendgrid/src/*"],
"@trigger.dev/airtable": ["../../integrations/airtable/src/index"],
"@trigger.dev/airtable/*": ["../../integrations/airtable/src/*"]
}
}
}
Creating a New Reference Project
This guide assumes that you have followed the Contributing.md instructions to set up a local trigger.dev instance. If not, please complete the setup before continuing.
Step-by-Step Instructions
- Run an HTTP tunnel: You will need to run an HTTP tunnel to expose your local webapp, it is required for some API calls during building the image to deploy on your local instance. This is optional if you do not plan to test deployment on your local instance.
- Download the ngrok CLI. This can be done by following the instructions on ngrok's website.
- Create an account on ngrok to obtain the authtoken and add it to the CLI.
ngrok config add-authtoken <your-auth-token>
Replace the with the token you obtain from ngrok.
- Run the tunnel.
ngrok http <your-app-port>
Replace the with the webapp port, default is 3030.
-
Add your tunnel URL to the env: After running the ngrok tunnel, you will see URL in your terminal, it will look something like
https://<your-tunnel-address>.ngrok-free.app. Replace theAPP_ORIGINvariable with this URL in your.envfile in the root of the trigger.dev project. -
Run the webapp on localhost:
pnpm run dev --filter webapp --filter coordinator --filter docker-provider
- Build the CLI in a new terminal window:
# Build the CLI
pnpm run build --filter trigger.dev
# Make it accessible to `pnpm exec`
pnpm i
- Set up a new project in the webapp:
- Open the webapp running on
localhost:3030. - Create a new project in the webapp UI.
- Go to the Project Settings page and copy the project reference id from there.
- Copy the hello-world project as a template:
cp -r references/hello-world references/<new-project>
Replace <new-project> with your desired project name.
- Update project details:
-
Open
<new-project>/package.jsonand change the name field. (Tip: Use the same name as in the webapp to avoid confusion.) -
Open
<new-project>/trigger.config.tsand update the project field with the project reference you copied from the webapp. -
Run
pnpm iin your<new-project>directory to sync the dependencies.
- Authorize the CLI for your project:
pnpm exec trigger login -a http://localhost:3030 --profile local
- Run the new project: You can now run your project using the CLI with the following command:
pnpm exec trigger dev --profile local
You can also deploy them against your local instance with the following command:
pnpm exec trigger deploy --self-hosted --load-image --profile local