
Research
/Security News
GlassWASM: WebAssembly Malware Found in Trojanized Open VSX Extensions
The trojanized extensions use TinyGo-compiled WebAssembly and Solana transaction memos to resolve command-and-control infrastructure.
@loopstack/custom-tool-example-module
Advanced tools
A complete example demonstrating how to implement and use a custom tool in a workflow
A module for the Loopstack AI automation framework.
This module provides a complete example demonstrating how to implement and use custom tools in a Loopstack workflow.
Custom tools are the building blocks of Loopstack automations. This module serves as a hands-on reference for developers learning how to extend Loopstack with their own functionality.
By exploring this example, you'll understand:
BaseTool with a handle() method@Tool decoratorwait: true on transitions for manual triggersThis is a great starting point before building your own custom tools.
npm install @loopstack/custom-tool-example-module
Then register the module in your app:
import { StudioApp } from '@loopstack/common';
import { CustomToolExampleWorkflow, CustomToolModule } from '@loopstack/custom-tool-example-module';
@StudioApp({
title: 'Custom Tool Example',
workflows: [CustomToolExampleWorkflow],
})
@Module({
imports: [CustomToolModule],
})
export class MyAppModule {}
Register the workflow, tools, and any supporting services as NestJS providers:
@Module({
providers: [CustomToolExampleWorkflow, MathSumTool, CounterTool, MathService],
exports: [CustomToolExampleWorkflow, MathSumTool, CounterTool, MathService],
})
export class CustomToolModule {}
A simple tool that maintains internal state across calls. It extends BaseTool and implements handle():
import { BaseTool, Tool, ToolResult } from '@loopstack/common';
@Tool({
name: 'counter',
uiConfig: {
description: 'Counter tool.',
},
})
export class CounterTool extends BaseTool<object, object, number> {
count: number = 0;
protected async handle(_args?: object): Promise<ToolResult<number>> {
this.count++;
return Promise.resolve({ data: this.count });
}
}
The count property persists across calls within the same workflow execution, so each call increments the counter.
A tool that accepts typed arguments via a Zod schema and uses NestJS constructor injection for services:
import { z } from 'zod';
import { BaseTool, Tool, ToolResult } from '@loopstack/common';
import { MathService } from '../services/math.service';
const MathSumSchema = z
.object({
a: z.number(),
b: z.number(),
})
.strict();
type MathSumArgs = z.infer<typeof MathSumSchema>;
@Tool({
name: 'math_sum',
uiConfig: {
description: 'Math tool calculating the sum of two arguments by using an injected service.',
},
schema: MathSumSchema,
})
export class MathSumTool extends BaseTool<MathSumArgs, object, number> {
constructor(private readonly mathService: MathService) {
super();
}
protected async handle(args: MathSumArgs): Promise<ToolResult<number>> {
const sum = this.mathService.sum(args.a, args.b);
return Promise.resolve({ data: sum });
}
}
The schema option on @Tool validates incoming arguments. The injected MathService is a standard NestJS injectable:
@Injectable()
export class MathService {
public sum(a: number, b: number) {
return a + b;
}
}
The workflow extends BaseWorkflow with typed arguments and typed state. The argument schema is defined in the @Workflow decorator:
interface CustomToolExampleState {
total?: number;
}
@Workflow({
uiConfig: __dirname + '/custom-tool-example.ui.yaml',
schema: z
.object({
a: z.number().default(1),
b: z.number().default(2),
})
.strict(),
})
export class CustomToolExampleWorkflow extends BaseWorkflow<{ a: number; b: number }, CustomToolExampleState> {
constructor(
private readonly counterTool: CounterTool,
private readonly mathTool: MathSumTool,
@Inject(DOCUMENT_STORE) private readonly documentStore: DocumentStore,
) {
super();
}
}
Tools registered in the same NestJS module are injected via the constructor. Call them with this.tool.call(args). call() is provided by BaseTool and delegates to your handle() implementation.
Call tools inside transition methods and store results in workflow state:
@Transition({ to: 'waiting_for_user' })
async calculate(
ctx: WorkflowContext,
args: { a: number; b: number },
state: CustomToolExampleState,
): Promise<CustomToolExampleState> {
const calcResult = await this.mathTool.call({ a: args.a, b: args.b });
const total = calcResult.data as number;
await this.documentStore.save(MessageDocument, {
role: 'assistant',
text: `Tool calculation result:\n${args.a} + ${args.b} = ${total}`,
});
await this.documentStore.save(MessageDocument, {
role: 'assistant',
text: `Alternatively, using workflow method:\n${args.a} + ${args.b} = ${this.sum(args.a, args.b)}`,
});
return { ...state, total };
}
The workflow then prints a second message using this.sum(args.a, args.b). That private helper is intentional demo redundancy. It shows the same calculation can also live as workflow-local logic. In production you would normally pick one approach: call a tool (reusable, injectable, schema-validated) or use a private helper (simple, workflow-specific). Here both are shown side by side for comparison. Note that total stored in workflow state comes from the tool result, not from sum().
The counter tool increments on each call, demonstrating that tool state persists within a workflow execution:
const c1 = await this.counterTool.call();
const c2 = await this.counterTool.call();
const c3 = await this.counterTool.call();
await this.documentStore.save(MessageDocument, {
role: 'assistant',
text: `Counter before pause: ${c1.data}, ${c2.data}, ${c3.data}\n\nPress Next to continue...`,
});
Use wait: true on a transition to pause the workflow until it is manually triggered (e.g., by a UI button):
@Transition({ from: 'waiting_for_user', to: 'resumed', wait: true })
async userContinue(ctx: WorkflowContext, state: CustomToolExampleState): Promise<CustomToolExampleState> {
return state;
}
The workflow pauses at waiting_for_user until the user triggers the userContinue transition.
A terminal @Transition method can return data as the workflow output:
@Transition({ from: 'resumed', to: 'end' })
async continueCount(ctx: WorkflowContext, state: CustomToolExampleState): Promise<{ total: number | undefined }> {
const c4 = await this.counterTool.call();
const c5 = await this.counterTool.call();
const c6 = await this.counterTool.call();
await this.documentStore.save(MessageDocument, {
role: 'assistant',
text: `Counter after resume: ${c4.data}, ${c5.data}, ${c6.data}\n\nIf state persisted, this should be 4, 5, 6.`,
});
return { total: state.total };
}
After resuming, the counter continues from where it left off (4, 5, 6), demonstrating that tool state survives a wait pause.
Define private methods for reusable logic within the workflow. In this example, sum() mirrors what MathSumTool already does. It exists only to demonstrate that alternative:
private sum(a: number, b: number) {
return a + b;
}
Use helpers for workflow-specific formatting or glue logic; use tools when the logic should be shared, tested independently, or exposed to agents.
import { Inject } from '@nestjs/common';
import { z } from 'zod';
import { BaseWorkflow, DOCUMENT_STORE, Final, Initial, MessageDocument, Transition, Workflow } from '@loopstack/common';
import type { DocumentStore, WorkflowContext } from '@loopstack/common';
import { CounterTool, MathSumTool } from '../tools';
interface CustomToolExampleState {
total?: number;
}
@Workflow({
uiConfig: __dirname + '/custom-tool-example.ui.yaml',
schema: z
.object({
a: z.number().default(1),
b: z.number().default(2),
})
.strict(),
})
export class CustomToolExampleWorkflow extends BaseWorkflow<{ a: number; b: number }, CustomToolExampleState> {
constructor(
private readonly counterTool: CounterTool,
private readonly mathTool: MathSumTool,
@Inject(DOCUMENT_STORE) private readonly documentStore: DocumentStore,
) {
super();
}
@Transition({ to: 'waiting_for_user' })
async calculate(
ctx: WorkflowContext,
args: { a: number; b: number },
state: CustomToolExampleState,
): Promise<CustomToolExampleState> {
const calcResult = await this.mathTool.call({ a: args.a, b: args.b });
const total = calcResult.data as number;
await this.documentStore.save(MessageDocument, {
role: 'assistant',
text: `Tool calculation result:\n${args.a} + ${args.b} = ${total}`,
});
await this.documentStore.save(MessageDocument, {
role: 'assistant',
text: `Alternatively, using workflow method:\n${args.a} + ${args.b} = ${this.sum(args.a, args.b)}`,
});
const c1 = await this.counterTool.call();
const c2 = await this.counterTool.call();
const c3 = await this.counterTool.call();
await this.documentStore.save(MessageDocument, {
role: 'assistant',
text: `Counter before pause: ${c1.data}, ${c2.data}, ${c3.data}\n\nPress Next to continue...`,
});
return { ...state, total };
}
@Transition({ from: 'waiting_for_user', to: 'resumed', wait: true })
async userContinue(ctx: WorkflowContext, state: CustomToolExampleState): Promise<CustomToolExampleState> {
return state;
}
@Transition({ from: 'resumed', to: 'end' })
async continueCount(ctx: WorkflowContext, state: CustomToolExampleState): Promise<{ total: number | undefined }> {
const c4 = await this.counterTool.call();
const c5 = await this.counterTool.call();
const c6 = await this.counterTool.call();
await this.documentStore.save(MessageDocument, {
role: 'assistant',
text: `Counter after resume: ${c4.data}, ${c5.data}, ${c6.data}\n\nIf state persisted, this should be 4, 5, 6.`,
});
return { total: state.total };
}
private sum(a: number, b: number) {
return a + b;
}
}
This workflow uses the following Loopstack modules:
@loopstack/common: Base classes, decorators, BaseTool, DocumentStore, and MessageDocumentAuthor: Jakob Klippel
License: MIT
FAQs
A complete example demonstrating how to implement and use a custom tool in a workflow
We found that @loopstack/custom-tool-example-module demonstrated a healthy version release cadence and project activity because the last version was released less than a year ago. It has 1 open source maintainer collaborating on the project.
Did you know?

Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.

Research
/Security News
The trojanized extensions use TinyGo-compiled WebAssembly and Solana transaction memos to resolve command-and-control infrastructure.

Security News
Anthropic says the directive cited national security concerns over a narrow jailbreak, but offered no specific technical details.

Security News
A network of 152 Chrome live wallpaper extensions hid ad tracking and made extension-driven traffic look like Google search clicks.