feat(apps): split AI tool and workflow action triggers in LogicFunction manifest (#20208)
## Summary Replaces the bolted-on `isTool` + `toolInputSchema` fields on `LogicFunctionManifest` with two distinct, opt-in triggers that align with the existing `cron` / `databaseEvent` / `httpRoute` trigger pattern: - **`toolTriggerSettings`** — exposes the function as an AI tool (chat / MCP / function calling). Uses standard JSON Schema (the format LLMs natively understand). - **`workflowActionTriggerSettings`** — exposes the function as a step in the visual workflow builder. Uses Twenty's rich `InputSchema` so the builder can render proper `FieldMetadataType`-aware editors, variable pickers, labels, and an optional `outputSchema`. A function can opt into none, one, or both. Each surface gets the schema format appropriate for it. ### Why `isTool: true` previously exposed the function as both an AI tool AND a workflow node, with the same JSON Schema feeding both — but the workflow builder really wants Twenty's `InputSchema` (with `CURRENCY`, `RELATION`, `EMAILS`, etc.) and the AI surface really wants standard JSON Schema. Today the workflow builder hacks around this by treating JSON Schema as `InputSchema`, which silently breaks for any non-primitive field type. Splitting the triggers fixes that and lets each surface evolve independently. ### Migration - **Fast** instance command adds the two new nullable columns. - **Slow** instance command backfills `toolTriggerSettings` + `workflowActionTriggerSettings` from `isTool=true` rows (preserving today's both-surfaces behaviour) then drops the legacy columns. ### Stacked Stacked on top of #20181. Merge that first, then this. ## Test plan - [ ] CI green (oxlint, typecheck, jest, vitest) - [ ] Run `--include-slow` upgrade against a workspace with existing `isTool=true` logic functions; verify both new columns populated and old columns dropped - [ ] Verify AI chat sees migrated tool functions (Linear create-issue, Exa search) and can call them with the JSON Schema - [ ] Add an AI-tool function from the Settings UI (toggles `toolTriggerSettings`) and verify it shows up in chat - [ ] Add a workflow-action function from the Settings UI (toggles `workflowActionTriggerSettings`) and verify it appears in the workflow node picker - [ ] In the workflow builder, edit a `LOGIC_FUNCTION` step and verify input fields render (no more JSON-Schema-as-InputSchema hack) - [ ] Try defining a function with no triggers in the SDK and verify `defineLogicFunction` rejects it 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: martmull <martmull@hotmail.fr>
This commit is contained in:
@@ -142,11 +142,14 @@ const handler = async (event: RoutePayload) => {
|
||||
Header names are normalized to lowercase. Access them using lowercase keys (e.g., `event.headers['content-type']`).
|
||||
</Note>
|
||||
|
||||
#### Exposing a function as a tool
|
||||
#### Exposing a function as an AI tool or workflow action
|
||||
|
||||
Logic functions can be exposed as **tools** for AI agents and workflows. When marked as a tool, a function becomes discoverable by Twenty's AI features and can be used in workflow automations.
|
||||
Logic functions can be exposed on two surfaces, each with its own trigger:
|
||||
|
||||
To mark a logic function as a tool, set `isTool: true`:
|
||||
- **`toolTriggerSettings`** — makes the function discoverable by Twenty's AI features (chat, MCP, function calling). Uses standard JSON Schema, the format LLMs natively understand.
|
||||
- **`workflowActionTriggerSettings`** — makes the function appear as a step in the visual workflow builder. Uses Twenty's rich `InputSchema` so the builder can render proper field editors, variable pickers, and labels.
|
||||
|
||||
A function can opt into one, the other, or both. They sit alongside `cronTriggerSettings`, `databaseEventTriggerSettings`, and `httpRouteTriggerSettings` — same pattern, same shape.
|
||||
|
||||
```ts src/logic-functions/enrich-company.logic-function.ts
|
||||
import { defineLogicFunction } from 'twenty-sdk/define';
|
||||
@@ -176,31 +179,33 @@ export default defineLogicFunction({
|
||||
description: 'Enrich a company record with external data',
|
||||
timeoutSeconds: 10,
|
||||
handler,
|
||||
isTool: true,
|
||||
toolTriggerSettings: {},
|
||||
});
|
||||
```
|
||||
|
||||
Key points:
|
||||
|
||||
- You can combine `isTool` with triggers — a function can be both a tool (callable by AI agents) and triggered by events at the same time.
|
||||
- **`toolInputSchema`** (optional): A JSON Schema object describing the parameters your function accepts. The schema is computed automatically from source code static analysis, but you can set it explicitly:
|
||||
- A function can mix surfaces — declare both `toolTriggerSettings` and `workflowActionTriggerSettings` to expose it in chat AND in the workflow builder.
|
||||
- `toolTriggerSettings.inputSchema` and `workflowActionTriggerSettings.inputSchema` are both optional. When omitted, the manifest builder infers them from the handler source code (JSON Schema for the AI tool, Twenty's `InputSchema` for the workflow action). Provide one explicitly when you want richer typing — for example, with `FieldMetadataType`-aware fields like `CURRENCY` or `RELATION` for the workflow builder, or with `description` fields the AI agent can read:
|
||||
|
||||
```ts
|
||||
export default defineLogicFunction({
|
||||
...,
|
||||
toolInputSchema: {
|
||||
type: 'object',
|
||||
properties: {
|
||||
companyName: {
|
||||
type: 'string',
|
||||
description: 'The name of the company to enrich',
|
||||
},
|
||||
domain: {
|
||||
type: 'string',
|
||||
description: 'The company website domain (optional)',
|
||||
toolTriggerSettings: {
|
||||
inputSchema: {
|
||||
type: 'object',
|
||||
properties: {
|
||||
companyName: {
|
||||
type: 'string',
|
||||
description: 'The name of the company to enrich',
|
||||
},
|
||||
domain: {
|
||||
type: 'string',
|
||||
description: 'The company website domain (optional)',
|
||||
},
|
||||
},
|
||||
required: ['companyName'],
|
||||
},
|
||||
required: ['companyName'],
|
||||
},
|
||||
});
|
||||
```
|
||||
@@ -239,7 +244,7 @@ yarn twenty exec --postInstall
|
||||
```
|
||||
|
||||
Key points:
|
||||
- Post-install functions use `definePostInstallLogicFunction()` — a specialized variant that omits trigger settings (`cronTriggerSettings`, `databaseEventTriggerSettings`, `httpRouteTriggerSettings`, `isTool`).
|
||||
- Post-install functions use `definePostInstallLogicFunction()` — a specialized variant that omits trigger settings (`cronTriggerSettings`, `databaseEventTriggerSettings`, `httpRouteTriggerSettings`, `toolTriggerSettings`, `workflowActionTriggerSettings`).
|
||||
- The handler receives an `InstallPayload` with `{ previousVersion?: string; newVersion: string }` — `newVersion` is the version being installed, and `previousVersion` is the version that was previously installed (or `undefined` on a fresh install). Use these values to distinguish fresh installs from upgrades and to run version-specific migration logic.
|
||||
- **When the hook runs**: on fresh installs only, by default. Pass `shouldRunOnVersionUpgrade: true` if you also want it to run when the app is upgraded from a previous version. When omitted, the flag defaults to `false` and upgrades skip the hook.
|
||||
- **Execution model — async by default, sync opt-in**: the `shouldRunSynchronously` flag controls *how* post-install is executed.
|
||||
|
||||
Reference in New Issue
Block a user