feat(llm): 支持增量上下文,为模型加 store: true 即走 relay

- 新增 instance.js:data_root/instance-id,首次启动生成 12 位 hex,跨 core 唯一
- engine 传 context_id = <core_id>:<user>:<session_id>,三层定位不撞车
- callModel 对 store: true 的模型只发新增消息 + X-Context-Id/Hash,
  409 时改发全量重试一次;指针存内存,重启自动回落全量
- config.js 把 store 字段透传到模型对象(以前会被丢掉)
- 未配 store 的模型行为完全不变
This commit is contained in:
2026-11-17 20:29:16 +08:00
parent 05d03f0836
commit bba42ae2ca
5 changed files with 87 additions and 2 deletions
+4
View File
@@ -3,6 +3,7 @@
const fs = require('fs');
const path = require('path');
const { buildMessages, callModel, encodeBlocks, decodeBlocks, compressMessages, titleMessages, setProfilePrompts, patchImageReceipt } = require('./llm');
const { instanceId } = require('./instance');
const { executeTool, resolveInTask } = require('./tools/executor');
const publicFiles = require('./public-files');
const mcp = require('./mcp');
@@ -1159,6 +1160,9 @@ async function streamAssistant(sessionId, run, session, messages, tools) {
// key_slot 定位会话占用的具体 key 副本(model_id 只到「平台/模型」)
const result = await callModel(cfg, session.model_id, messages, tools, {
key_slot: session.model_slot,
// 增量上下文用:<core_id>:<user>:<session_id>,三层保证跨 core / 跨用户 / 跨会话都不撞。
// 只有带 store: true 的模型才会真的用它(llm.js 判断)。
context_id: instanceId(cfg.data_root) + ':' + (session.user || '-') + ':' + session.id,
signal: run.abortController.signal,
shouldCancel: () => run.cancelled,
onText: delta => bus.emit(sessionId, 'assistant_delta', { text: delta }),