ModelMessage

CORE · interface · Type contract

One provider-neutral turn; images are user visual input and are not part of the text budget.

Import

TypeScript

import type { ModelMessage } from '@jangts/agent-core';

JavaScript

/** @typedef {import('@jangts/agent-core').ModelMessage} ModelMessage */

Other public entry points:@jangts/agent-core, @jangts/agent-core/types, @jangts/agent-core/structs.

Definition

interface ModelMessage

Field structure

Field	Type	Required	Description
role	"system" | "user" | "assistant" | "tool"	Required	See the type definition and usage context.
content	string	Required	Rendered text content sent to the model.
images	readonly ModelImage[] | undefined	Optional	User-supplied visual input. Binary payload is separate from the text character budget.
toolCalls	readonly ModelToolCall[] | undefined	Optional	Tool declarations made by an assistant turn; rejected on every other role at runtime.
toolCallId	string | undefined	Optional	Call id answered by a tool result; rejected on every other role at runtime.

Examples

• Process observation and node reports: The App subscribes to Core execution facts; nodes report their own usedModel and use an instance sub-logger to record intent first and then results, while the Loop owns identity and logging.
• ModelRunner: input can be shortened while the output contract stays the same: Execute one model interaction through the real Provider port and return a value plus the invocation for this call.
• Actual request measurement and same-call preparation snapshot: Use prepare only when context must be handled ahead of time; multiple measurements and the final send reuse the same transport preparation.
• Model tools and transports: typed selection only after full validation: A tool definition provides a Schema and resolver; a transport translates wire output into the unified model selection contract.
• Compatible Provider and custom Providers: Support owns the common Chat Completions HTTP/SSE lifecycle; an App can substitute its own ChatProvider.
• Context fragments: explicit source, trust, and budget: Core recognizes fragments and the assembly contract, validates source metadata and budgets, and has no built-in business compression algorithm.
• Compressors and selectors: recommended implementations and custom algorithms: Bind prompts, output parsing, and algorithms as handlers; model I/O goes through callModel of the same request.
• Support: the aggregate entry for recommended usage: Support aggregates recommended construction, policies, and same-source configuration generation; Core execution does not depend on this namespace.

Observe the nodes and model events a task passes through

Beginner · Node.js

TypeScript

import { AgentCoreEngine, defineWorkflow, route } from '@jangts/agent-core/engine';

type State = { name: string; answer?: string };
const workflow = defineWorkflow<State, string>({
  id: 'greeting',
  nodes: [{
    id: 'greet', description: '根据本次姓名生成问候', startsRound: true,
    run: state => ({ answer: '你好，' + state.name }),
    routes: [route.deliver('answer')],
  }],
});
const agent = new AgentCoreEngine({ workflow, execution: {
  observe: { event(event) { console.log(event.type, event.eventId); } },
} });
console.log(await agent.run({ name: 'Ivan' }));

JavaScript

import { AgentCoreEngine, defineWorkflow, route } from '@jangts/agent-core/engine';
const workflow = defineWorkflow({
    id: 'greeting',
    nodes: [{
            id: 'greet', description: '根据本次姓名生成问候', startsRound: true,
            run: state => ({ answer: '你好，' + state.name }),
            routes: [route.deliver('answer')],
        }],
});
const agent = new AgentCoreEngine({ workflow, execution: {
        observe: { event(event) { console.log(event.type, event.eventId); } },
    } });
console.log(await agent.run({ name: 'Ivan' }));

Report a business stage while the node executes

Configuration · Node.js

The node only reports facts it knows; Core owns node identity and execution association.

TypeScript

import { AgentCoreEngine, defineWorkflow, route } from '@jangts/agent-core/engine';
type State = { name: string; answer?: string };
const workflow = defineWorkflow<State, string>({ id: 'greeting', nodes: [{
  id: 'greet', description: '生成问候', startsRound: true,
  run(state, context) {
    context.report({ payload: { phase: 'formatting' } });
    return { answer: '你好，' + state.name };
  }, routes: [route.deliver('answer')],
}] });
const agent = new AgentCoreEngine({ workflow, execution: {
  observe: { event(event) { console.log(event.type); } },
} });
console.log(await agent.run({ name: 'Ivan' }));

JavaScript

import { AgentCoreEngine, defineWorkflow, route } from '@jangts/agent-core/engine';
const workflow = defineWorkflow({ id: 'greeting', nodes: [{
            id: 'greet', description: '生成问候', startsRound: true,
            run(state, context) {
                context.report({ payload: { phase: 'formatting' } });
                return { answer: '你好，' + state.name };
            }, routes: [route.deliver('answer')],
        }] });
const agent = new AgentCoreEngine({ workflow, execution: {
        observe: { event(event) { console.log(event.type); } },
    } });
console.log(await agent.run({ name: 'Ivan' }));

Related policy source

• Shared execution fact source and consumption isolation
• Sub-step statuses and their progression
• Core default budget values
• Same-call preparation and real input measurement
• HTTP failure mapping and controlled request fields
• Fragment validation and budget rejection
• Default selection instruction and excerpt range
• Compression request and the real callModel contract

Related definitions

• ModelImage
• ModelToolCall