PreparedModelCall

CORE · interface · 类型合同

Request-bound, single-use transport snapshot.

Repeated measurements do not prepare or spend calls.

导入

TypeScript

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

JavaScript

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

其他公开入口：@jangts/agent-core、@jangts/agent-core/model、@jangts/agent-core/types、@jangts/agent-core/structs。

定义

interface PreparedModelCall<Result>

继承或实现： PreparedModelTransport<Result>。

泛型参数

类型参数	约束	默认类型
Result	—	unknown

字段结构

字段	类型	必需	说明
modelId	string | undefined	可省略	只读。Model binding the call was prepared with.
stageId	string	必需	只读。见类型定义及调用场景。
transportId	string	必需	只读。见类型定义及调用场景。
maxOutputTokens	number	必需	只读。见类型定义及调用场景。
remainingCalls	number	必需	只读。Calls still available; a live value, so later spends are visible.
measure	(input: string | readonly ModelMessage[] | BuiltContext) => ModelInputMeasurement	必需	Preflight measurement of one input shape. Never marks the call used and never charges the budget.
options	Pick<ModelOptions, "functions" | "json" | "toolChoice" | "parallelToolCalls" | "temperature">	必需	只读。见类型定义及调用场景。
schemaChars	number	必需	只读。见类型定义及调用场景。
instruction	string | undefined	可省略	只读。见类型定义及调用场景。
createResponse	() => ModelResponseCollector<Result>	必需	见类型定义及调用场景。

成员目录

成员	用途
measure	Preflight measurement of one input shape. Never marks the call used and never charges the budget.
createResponse	完整输入输出见下方重载。

成员详情

measure

Preflight measurement of one input shape.

Never marks the call used and never charges the budget.

PreparedModelCall.measure(input: string | readonly ModelMessage[] | BuiltContext): ModelInputMeasurement

Preflight measurement of one input shape.

Never marks the call used and never charges the budget.

参数	类型	传递	说明
input	string | readonly ModelMessage[] | BuiltContext	必需	—

参数 input的可选结构： 选择其中一种结构，各分支的字段要求分别列出。

分支 1

string

分支 2

readonly ModelMessage[]

分支 3

字段	类型	必需	说明
messages	ModelMessage[]	必需	Messages in final model order.
fragmentUsage	ContextFragmentUsage[]	必需	Per-fragment accounting in the same order as the fragments were merged.
inputChars	number	必需	Total counted characters across messages and usage.
maxInputChars	number	必需	Budget the assembly was validated against; Infinity means unbounded.

返回： ModelInputMeasurement。

返回值结构

字段	类型	必需	说明
messageChars	number	必需	只读。Characters of the message content.
instructionChars	number	必需	只读。Characters of the instruction text.
schemaChars	number	必需	只读。Characters of the compiled capability schemas.
transportChars	number	必需	只读。Characters the transport itself contributes (for example framing).
inputChars	number	必需	只读。Sum of the character components above.
maxInputChars	number	必需	只读。Character budget the input was compared against.
maxSchemaChars	number	必需	只读。Schema budget the input was compared against.
maxOutputTokens	number	必需	只读。Output tokens reserved for this call, which reduces the token input ceiling.
counterId	string | undefined	可省略	只读。Tokenizer identity; present only when a token budget applies.
inputTokens	number | undefined	可省略	只读。Measured input tokens.
inputTokenLimit	number | undefined	可省略	只读。Token ceiling after reserving output and the safety margin.
withinBudget	boolean	必需	只读。True when every applicable character and token budget accepted the input.

createResponse

PreparedModelCall.createResponse(): ModelResponseCollector<Result>

无调用参数。

返回： ModelResponseCollector<Result>。

返回值结构

字段	类型	必需	说明
call	{ id: string; name: string; argumentChars: number; } | undefined	可省略	只读。Native function call that carried the selection, when the response used one.
calls	readonly { id: string; name: string; argumentChars: number; toolId?: string; }[] | undefined	可省略	只读。All native calls carried by a session response; omitted by single-selection collectors.
outputChars	number	必需	只读。Output characters accumulated so far, for output-budget accounting.
push	(part: ModelPart) => void	必需	Appends one streamed part; parts arrive in provider order.
finish	() => Result	必需	Finalizes the response. Throws when the accumulated content is not a valid selection.

使用示例

• 实际请求测量与同次准备快照：需要提前处理上下文时才 prepare；多次测量与最终发送复用同一次传输准备。

普通调用仍然只有 invoke

入门 · Node.js

TypeScript

import { ModelRunner } from '@jangts/agent-core/model';
const provider = {
  async *stream(messages: readonly { content: string }[]) {
    yield '收到：' + messages.at(-1)!.content;
  },
};
const runner = new ModelRunner({ provider });
console.log((await runner.invoke('你好')).value);

JavaScript

import { ModelRunner } from '@jangts/agent-core/model';
const provider = {
    async *stream(messages) {
        yield '收到：' + messages.at(-1).content;
    },
};
const runner = new ModelRunner({ provider });
console.log((await runner.invoke('你好')).value);

只在需要测量时准备一次

配置 · Node.js

TypeScript

import { ModelRunner } from '@jangts/agent-core/model';
const provider = {
  async *stream(messages: readonly { content: string }[]) {
    yield '收到：' + messages.at(-1)!.content;
  },
};
const runner = new ModelRunner({ provider, maxInputChars: 1000 });
const prepared = runner.prepare({ stageId: 'reply', maxOutputTokens: 400 });
const input = '你好，解释这次服务状态';
console.log(prepared.measure(input));
const result = await runner.invoke({ input, prepared });
console.log(result.value);

JavaScript

import { ModelRunner } from '@jangts/agent-core/model';
const provider = {
    async *stream(messages) {
        yield '收到：' + messages.at(-1).content;
    },
};
const runner = new ModelRunner({ provider, maxInputChars: 1000 });
const prepared = runner.prepare({ stageId: 'reply', maxOutputTokens: 400 });
const input = '你好，解释这次服务状态';
console.log(prepared.measure(input));
const result = await runner.invoke({ input, prepared });
console.log(result.value);

相关策略源码

• 同次准备与真实输入计量

相关定义

• BuiltContext
• ModelInputMeasurement
• ModelMessage
• ModelOptions
• ModelResponseCollector
• PreparedModelTransport