feat: v0.7.3 成本收口 · 状态一致 · 死账清理 — Prompt Cache 根治 + SSRF DNS Pinning + 87 用例扩充全量回归
P1 修复面收口: Prompt Cache 根治(日期/记忆/附件三类易变内容出 system 入用户消息 前置块 user-context.ts, system 跨 run 字节级稳定; Anthropic system 块数组化 + cache_control ephemeral 断言, DeepSeek 自动缓存前缀命中 — 多轮对话输入 token 成本降数量级); 编辑重发/重新生成幽灵 Trace 双侧根治(DB truncateMessagesAfter 同步过滤 metadata.traceSteps + 前端 trimTraceStepsByAnchor 镜像, 严格小于锚点 时间戳, 同毫秒等值判废); sessions:deleteMessage 死通道全链路删除(渲染层零调用 + message_count 漂移面); Ollama vision 能力门控全链路(MetonaModelInfo .supportsVision 贯穿 adapter/IPC/store/UI, model-capabilities.ts 三道判定纯函数, 未知保守放行); 记忆固化节流(consolidation-policy 纯函数: 总开关 + 内容门控 [回答>=200字符或存在成功工具调用] + 会话级 10 分钟频率窗口, 三 memory.* 配置键) P2 安全纵深: SSRF DNS Pinning 关闭 rebinding 窗口(ssrf-guard 重构 resolvePublicAddresses 单源; ssrf-dispatcher 以 undici Agent.connect.lookup 钉死校验 IP, TLS SNI 保持原域名, 一次性 dispatcher 用后即毁; 代理激活显式 退化为仅入口校验); web_fetch 重写手动逐跳重定向循环(每跳先校验后连接, 替代 redirect:follow 内核跟跳的中间跳裸奔, 上限 5 跳); http_request 换用 pinned fetch; web_search 可达性预检加固(私有 URL 零请求 + 不跟跳, 3xx 视为 可达); Agent 浏览器 CORS 通配收紧为 Origin 回显 + Vary: Origin; ConfirmationHook.forgetSession 会话终态清理(会话删除/abort 联动/SubAgent 终结三处接线, 根治 rememberedDecisions 泄漏) P3 架构还债: agent.enableReflection 死配置全链路接线(main→shared→引擎→ Orchestrator→设置开关, REFLECTING 状态真实可达); AgentLoopConfig.timeoutMs 死字段删除; MemoryManager.cleanupExpired 挂入健康检查周期(expires_at 回收 管道真实化); buildSafeEnv 收敛 utils/safe-env.ts 单源(run_command 与 MCP stdio 共用, 终结双实现漂移); Trace 生命周期治理(metadata 只保留最近 20 个 run — keepRecentRuns 纯函数; JSONL 录制启动自动清理保留 200 个 + 设置页 手动清理); SLO/健康快照可视化(app:healthSnapshot IPC + 设置页只读卡片 + 审计链一键校验) P4 能力演进: 会话标题 LLM 自动生成(TitleGenerator — 每会话幂等/并发重入复用 同一 Promise/自定义标题不覆盖/失败静默回退, Sidebar 经 config:changed 实时 刷新); MCP 自动重连(5s/15s/60s 退避最多 3 次, reconnecting 状态机, teardownConnection 内部拆除保留簿记 — 用户断开/开关关闭即时取消, 设置页 显示第 N/3 次); 死循环检测 ABAB 乒乓模式(最近4轮 A→B→A→B 交替判定, 补齐 docs 第五章"两状态反复切换"检测契约); i18n 第三阶段(ChatInput/LLMSettings/ OnboardingWizard/MemoryViewer 主链路文案出层, zh-CN + en-US 双字典补齐) 测试: 737 → 824 用例(+87, 新增 8 个测试文件 + 扩展 3 个)。新覆盖: user-context 分组/空值收缩/拼接契约、context-builder 字节级稳定性、Anthropic cache_control 四态、consolidation-policy 九路判定矩阵、ssrf-dispatcher(pinned lookup/重定向 解析/IP 校验)、forget-session 会话隔离、trace-lifecycle run 淘汰、 trace-trim 严格小于边界、safe-env 净化矩阵、mcp-reconnect 退避状态机 (fake timers)、title-generator 并发重入、SQLite 侧 truncate×TRACE 联动 (Electron ABI)。测试驱动修复: GIT_*/ 注释终止块注释、重连计数被自身重试 前置断开重置(拆 teardownConnection 保留簿记)、TitleGenerator 幂等占位与 并发去重的检查顺序竞态(去重先于幂等) 版本: 0.7.3; README 同步(配置表新增 agent.enableReflection/memory.*/mcp.autoReconnect) 回归: typecheck 双端 0 错误; ESLint 0/0; 系统 Node 771 通过 53 跳过 (better-sqlite3 ABI); Electron ABI 全量 824/824 零跳过
This commit is contained in:
+66
-36
@@ -32,6 +32,7 @@ import { WindowManager } from './services/window-manager.service';
|
||||
import { MCPManager } from './services/mcp-manager.service';
|
||||
import { SessionSummaryService } from './services/session-summary.service';
|
||||
import { AgentEngineManager } from './services/agent-engine-manager.service';
|
||||
import { TitleGenerator } from './services/title-generator.service';
|
||||
import { ContextBuilder } from './harness/prompts/context-builder';
|
||||
import { MemoryManager } from './harness/memory/manager';
|
||||
import { MemoryConsolidator } from './harness/memory/consolidator';
|
||||
@@ -317,6 +318,8 @@ async function initialize(): Promise<void> {
|
||||
const traceEnabled = configService.get<boolean>('logging.traceEnabled') !== false;
|
||||
// SessionRecorder 录制总开关(false 时不写 TRACE JSONL 文件)
|
||||
sessionRecorder.setEnabled(traceEnabled);
|
||||
// v0.7.3 P3-3: 启动期清理旧 JSONL 录制文件(按 mtime 保留最近 200 个)
|
||||
void sessionRecorder.pruneOldRecordings();
|
||||
|
||||
// ===== v0.2.0: 安全模块(必须在工具注册之前) =====
|
||||
const policyEngine = new PolicyEngine();
|
||||
@@ -426,6 +429,9 @@ async function initialize(): Promise<void> {
|
||||
| 'max'
|
||||
| null) ?? 'high',
|
||||
toolExecutionTimeoutMs: configService.get<number>('agent.toolExecutionTimeoutMs') ?? 120_000,
|
||||
// v0.7.3 P3-1 接线: agent.enableReflection 此前为死配置(引擎读取但全链路
|
||||
// 无置 true 路径)。现注入引擎基线配置,REFLECTING 状态由此开关真实驱动。
|
||||
enableReflection: configService.get<boolean>('agent.enableReflection') === true,
|
||||
// F-8 接通: llm.temperature / llm.maxTokens 此前为死配置(引擎硬编码 0.0/63488)
|
||||
temperature: configService.get<number>('llm.temperature') ?? 0.0,
|
||||
maxTokens: configService.get<number>('llm.maxTokens') ?? 63488,
|
||||
@@ -490,6 +496,8 @@ async function initialize(): Promise<void> {
|
||||
mcpManager.setOnToolsChanged(() => {
|
||||
agentEngineManager.setToolsAll(toolRegistry.listTools());
|
||||
});
|
||||
// v0.7.3 P4-2: MCP 自动重连开关(默认开启;配置变更经 shared.ts applyConfigSideEffects 联动)
|
||||
mcpManager.setAutoReconnect(configService.get<boolean>('mcp.autoReconnect') !== false);
|
||||
const toolsReadyRef: ToolsReadyRef = { ready: false, toolCount: 0 };
|
||||
mcpManager
|
||||
.initialize()
|
||||
@@ -542,6 +550,8 @@ async function initialize(): Promise<void> {
|
||||
| null) ?? 'high',
|
||||
contextLength: ollamaNumCtx ?? undefined,
|
||||
contextWindow: buildAdapter().getContextWindow(),
|
||||
// v0.7.3 P3-1: SubAgent 与主引擎同源消费 enableReflection
|
||||
enableReflection: configService.get<boolean>('agent.enableReflection') === true,
|
||||
},
|
||||
);
|
||||
toolRegistry.registerBuiltin(new DelegateTaskTool(orchestrator));
|
||||
@@ -647,6 +657,56 @@ async function initialize(): Promise<void> {
|
||||
}
|
||||
};
|
||||
|
||||
// ===== P1-12 + v0.7.3 P3-4: SLO 健康监控接入(提前至窗口创建前,
|
||||
// 使 IPC 侧 getHealthSnapshot 闭包可安全引用 sloMonitor/healthChecker) =====
|
||||
const healthChecker = new HealthChecker(
|
||||
() => db,
|
||||
join(workspaceInfo.path, '.metona', 'agent.db'),
|
||||
);
|
||||
const sloMonitor = new SLOMonitor();
|
||||
agentEngineManager.on('complete', (data: { durationMs: number; terminationReason?: string }) => {
|
||||
sloMonitor.recordRequest(data.durationMs, data.terminationReason === 'completed');
|
||||
});
|
||||
/** 最近一次健康检查报告(供 IPC app:healthSnapshot 展示;初始 null = 尚未检查) */
|
||||
let lastHealthReport: Awaited<ReturnType<HealthChecker['check']>> | null = null;
|
||||
const healthTimer = setInterval(async () => {
|
||||
try {
|
||||
// v0.7.3 P3-1 接线: cleanupExpired 此前零调用方(且 expires_at 无写入方)双重死路。
|
||||
// 现挂入健康检查周期,让过期记忆(episodic_memories.expires_at)真正可回收,
|
||||
// 为后续记忆 TTL 语义预留真实管道;无过期条目时为零开销(一条带索引的 DELETE)。
|
||||
const expired = memoryManager.cleanupExpired();
|
||||
if (expired > 0) {
|
||||
log.info(`[Health] Expired episodic memories purged: ${expired}`);
|
||||
}
|
||||
const report = await healthChecker.check();
|
||||
lastHealthReport = report;
|
||||
if (!report.healthy) {
|
||||
const failed = report.checks
|
||||
.filter((c) => !c.healthy)
|
||||
.map((c) => c.name)
|
||||
.join(', ');
|
||||
log.warn(`[Health] Unhealthy checks: ${failed}`);
|
||||
trayManager?.setStatus('error');
|
||||
} else {
|
||||
// 恢复 idle(运行中状态会被后续 stateChange 覆盖)
|
||||
trayManager?.setStatus('idle');
|
||||
}
|
||||
const slo = sloMonitor.getStatus();
|
||||
if (slo.violated) {
|
||||
log.warn(
|
||||
`[SLO] Burn rate ${slo.burnRate.toFixed(2)} exceeds budget (errorRate=${(slo.errorRate * 100).toFixed(1)}%, ` +
|
||||
`P95=${slo.percentiles['P95'] ?? 0}ms, ${slo.totalRequests} requests)`,
|
||||
);
|
||||
}
|
||||
} catch (err) {
|
||||
log.warn('[Health] Check failed:', err);
|
||||
}
|
||||
}, 60_000);
|
||||
healthTimer.unref?.();
|
||||
|
||||
// ===== v0.7.3 P4-1: 会话标题生成器(首轮 run 完成后 LLM 生成精炼标题) =====
|
||||
const titleGenerator = new TitleGenerator(() => agentEngineManager.getAdapter(), sessionService);
|
||||
|
||||
// ===== 窗口管理 =====
|
||||
windowManager = new WindowManager();
|
||||
// 局部 const 捕获:闭包内 TS 无法对模块级可空变量做流收窄
|
||||
@@ -682,6 +742,12 @@ async function initialize(): Promise<void> {
|
||||
memoryConsolidator,
|
||||
orchestrator,
|
||||
sessionSummaryService,
|
||||
titleGenerator,
|
||||
getHealthSnapshot: () => ({
|
||||
slo: sloMonitor.getStatus(),
|
||||
health: lastHealthReport,
|
||||
generatedAt: Date.now(),
|
||||
}),
|
||||
toolsReadyRef,
|
||||
});
|
||||
},
|
||||
@@ -733,42 +799,6 @@ async function initialize(): Promise<void> {
|
||||
},
|
||||
);
|
||||
|
||||
// ===== P1-12: SLO 健康监控接入(原为死代码,现真实运行) =====
|
||||
const healthChecker = new HealthChecker(
|
||||
() => db,
|
||||
join(workspaceInfo.path, '.metona', 'agent.db'),
|
||||
);
|
||||
const sloMonitor = new SLOMonitor();
|
||||
agentEngineManager.on('complete', (data: { durationMs: number; terminationReason?: string }) => {
|
||||
sloMonitor.recordRequest(data.durationMs, data.terminationReason === 'completed');
|
||||
});
|
||||
const healthTimer = setInterval(async () => {
|
||||
try {
|
||||
const report = await healthChecker.check();
|
||||
if (!report.healthy) {
|
||||
const failed = report.checks
|
||||
.filter((c) => !c.healthy)
|
||||
.map((c) => c.name)
|
||||
.join(', ');
|
||||
log.warn(`[Health] Unhealthy checks: ${failed}`);
|
||||
trayManager?.setStatus('error');
|
||||
} else {
|
||||
// 恢复 idle(运行中状态会被后续 stateChange 覆盖)
|
||||
trayManager?.setStatus('idle');
|
||||
}
|
||||
const slo = sloMonitor.getStatus();
|
||||
if (slo.violated) {
|
||||
log.warn(
|
||||
`[SLO] Burn rate ${slo.burnRate.toFixed(2)} exceeds budget (errorRate=${(slo.errorRate * 100).toFixed(1)}%, ` +
|
||||
`P95=${slo.percentiles['P95'] ?? 0}ms, ${slo.totalRequests} requests)`,
|
||||
);
|
||||
}
|
||||
} catch (err) {
|
||||
log.warn('[Health] Check failed:', err);
|
||||
}
|
||||
}, 60_000);
|
||||
healthTimer.unref?.();
|
||||
|
||||
// ===== 应用生命周期 =====
|
||||
app.on('window-all-closed', () => {
|
||||
// macOS: 保持应用运行(托盘模式)
|
||||
|
||||
Reference in New Issue
Block a user