AskaEth/Cyrene_v2
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
262 Commits 7 Branches 0 Tags
53002f1e11c0759705045b25f139f09d6cd9e818
Commit Graph
5 Commits
Author SHA1 Message Date
AskaEth 53002f1e11 feat(model): LLM token 预算可配置化 — models.json每模型maxTokens + LLM_FALLBACK_MAX_TOKENS回退兜底
CI / Backend (build / vet / test / cross-compile) (push) Has been cancelled
Details
CI / Frontend (tsc + vite build) (push) Has been cancelled
Details 
- config.go: LLMConfig.FallbackMaxTokens(LLM_FALLBACK_MAX_TOKENS, 默认8192)
- runtimeconfig/models.go: Group.MaxTokens(models.json每模型独立预算, 负值拒绝)
- model/fallback.go: Chain.WithTokenBudget + applyBudget; runtime.go buildSnapshot接入(active.MaxTokens, env.FallbackMaxTokens)
- 优先级: 显式Request.MaxTokens > 组maxTokens > 仅多步链末端的LLM_FALLBACK_MAX_TOKENS > 适配器默认8192; 单步链不动env兜底(避免OpenAI主模型行为回归)
- gateway admin: AdminLlmProvider.MaxTokens 透传/回显/负值400
- .env.example + .env(LLM_FALLBACK_MAX_TOKENS=8192); Docs/api-client v0.27 + api-reference + design/04
- 新增测试(config/runtimeconfig/budget_test/admin_write), build/vet/test绿+3平台交叉编译
 2026-09-08 22:42:27 +08:00
Aska e037fab7d0 feat(weixin): 微信iLink扫码登录——面板触发+二维码显示+状态轮询+确认热连接
CI / Backend (build / vet / test / cross-compile) (push) Has been cancelled
Details
CI / Frontend (tsc + vite build) (push) Has been cancelled
Details 
适配器: login_admin.go(StartLogin 幂等取二维码/LoginStatus idle-waiting-confirmed-failed);
       login.go 重写 Connect 委托 StartLogin+等待,后台 runLogin 轮询,completeLogin/failLogin
       (锁序 loginMu->a.mu 不嵌套反向无死锁); adapter.go 加登录会话字段
网关: admin_channels_weixin.go(2 端点 login-qr/login-status owner+审计, confirmed 合并写回
       channels.json 保留 base_url/cdn/黑白名单 + Reload 热重载连接; qrImage best-effort Base64,
       token 掩码) + server.go 路由
前端: WeixinLogin.tsx(token空→扫码按钮→二维码→1.5s轮询→confirmed提示已登录+显示bot_id),
       AdapterDetailPage 渲染, api/types 补类型
必需偏离(最小化): 允许空token的weixin账号被构建/持久化(channels.go+runtimeconfig移出
       tokenRequiringPlatforms),否则空token账号无法注册调用StartLogin; 空token保持断开(configured
       仍要求token非空),确认后才持久化连接——守红线。
测试: login_admin/admin端点用例; 后端 build/vet/test 全绿 + 前端 build 101模块绿.
 2026-09-07 20:50:28 +08:00
Aska b614633432 fix: KOOK 连不上根因——空token裸账号不注册/占名 + open指向带token账号 + 前端token不覆盖真值
CI / Backend (build / vet / test / cross-compile) (push) Has been cancelled
Details
CI / Frontend (tsc + vite build) (push) Has been cancelled
Details 
agent-119 后端:
- buildAccountAdapter: token平台(kook/telegram/weixin)BotToken空→跳过build(不注册不占名), channelAccountsFromEnv 无tokenenv不合成账号(杜绝裸kook id="")
- runtimeconfig cleanAccounts: LoadChannels/SaveChannels 丢弃token平台空token账号(清理channels.json现有裸kook, 下次保存抹掉)
- findChannel: 裸平台名(kook)回退到该平台第一个带token账号, open"kook"不再命中裸的missing token

我补 前端token掩码不就覆真值:
- openConfig 不再把'（已配置，仅掩码显示）'当token回填(setToken空)→save token空不写bot_token保留真值
- 加 tokenHint(显示'已配置留空不改'/重填换); 修之前channels.json kook:3fc8 token变掩码串的bug
全仓 build/vet/test绿+前端build绿(463行)
 2026-09-07 16:01:47 +08:00
Aska aedae2190b feat(config): 运行时配置体系②——渠道配置落 channels.json + 面板落盘 + 热更新
CI / Backend (build / vet / test / cross-compile) (push) Has been cancelled
Details
CI / Frontend (tsc + vite build) (push) Has been cancelled
Details 
- runtimeconfig/channels.go: channels.json 承载 map[platform][]ChannelAccount(token/baseURL/黑白名单/群聊限流), LoadChannels 优先文件/回退env, SaveChannels(0600+tmp原子写), Upsert/Delete/Masked脱敏
- channelbridge/reload.go: adapterManager 热替换——工厂重建新adapter先连接, 再原子断旧adapter(独立readCtx/进行中不中断); Bridge.Reload(platform) 逐账号增删
- 面板 PUT/DELETE /admin/channels/{platform}/config(owner+审计) 落盘channels.json+Reload热更新; GET /admin/channels 附账号掩码不泄裸token
- 前端 Channels.tsx 配置表单(id/token/baseURL/黑白名单/限流+config JSON)+保存热更新提示+掩码展示/删除
- .gitignore 加 channels.json; .env 退兜底(.env.example注明); token不打印/Config()脱敏
- channelbridge.go 拆 logic→reload.go(742→731); 全仓 go build/vet/test绿+前端build绿
- 未做真token端到端(需渠道凭据,不读.env)——如实标注待凭据
 2026-09-07 13:29:41 +08:00
Aska 2d7151f979 feat(config): 运行时配置体系①——LLM 多厂商独立组落 models.json + 面板落盘 + 热更新
CI / Backend (build / vet / test / cross-compile) (push) Has been cancelled
Details
CI / Frontend (tsc + vite build) (push) Has been cancelled
Details 
- runtimeconfig/models.go: models.json 承载 [{id,provider,baseUrl,apiKey,model,fallback?,multimodal?}] 多厂商独立组(首项=生效)
- LoadModels 优先文件/回退env(GroupFromEnv), SaveModels(0600+tmp原子写), Activate/Upsert/MaskedKey脱敏
- 面板 POST /admin/llm 落盘 models.json(取代temporary内存) + GET 列组(脱敏)
- runtime/model/runtime.go: RuntimeLLM(adapter+router) atomic快照可热替换, Reload 重建不中断进行中请求; per-role路由用接口
- wire 用 NewRuntimeLLM(adapter+router), 保留 a.modelAdapter(=runtime stack, 审查模型随热更新)
- .env 退兜底(.env.example注明); apiKey不打印/Config()脱敏/热更新不中断
- 全仓 go build/vet/test 绿(54包)+交叉编译+前端build绿(79模块); api-client v0.15已同步
- 遗留: vision ModelInfo 仍读静态env(后续纳入); 保存组=生效组(第一版语义)
 2026-09-07 13:07:58 +08:00