i/Aether
1
0
Fork 0
mirror of https://github.com/fawney19/Aether.git synced 2026-10-05 00:47:48 +08:00
Code Issues Packages Projects Releases Wiki Activity
2,811 Commits 6 Branches 182 Tags
ddcbeb3ae99e4145c2a56ad261ab1d73b1c6bfdb
Commit Graph
28 Commits
Author SHA1 Message Date
ZheFox 45c840b8d3 fix(providers): refresh Antigravity grouped quotas 2026-09-03 10:46:44 +08:00
ZheFox 4da8c57fe3 fix(ai): align Gemini and Responses compatibility 2026-08-27 22:13:20 +08:00
ZheFox b13d9b9b40 fix(codex): restore upstream model discovery 2026-08-15 19:36:28 +08:00
zhefox 810c3dfe2b fix(codex): serve versioned dynamic model catalogs 2026-08-14 18:41:44 +08:00
MMEXA dfa121dd5b feat(openai): align GPT-5.6 and Codex request contracts 2026-07-11 07:40:12 +08:00
MMEXA c7186e1720 完善 Antigravity 配额展示与 CI 断言 2026-07-08 22:34:29 +08:00
MMEXA 6c4e730e60 修复 Antigravity OAuth 配额复检缺 project 2026-06-19 22:21:22 +08:00
fawney19 b59c724455 Normalize endpoint API root handling 2026-05-29 02:29:33 +08:00
Mas0nShi 9ad9858ac2 Merge origin/main into fix/gemini-cli-v1internal 2026-05-28 11:58:00 +08:00
zhefox 1173a4d9d5 fix(provider): split partial model fetch warnings from errors 2026-05-26 17:45:06 +08:00
zhefox aa409a8a9c fix(provider): refine endpoint default paths for openai and claude roots 2026-05-26 16:50:36 +08:00
zhefox d28a389a93 fix(provider): support unversioned API roots in model fetch 2026-05-26 15:39:56 +08:00
Mas0nShi 8b7643e150 Merge remote-tracking branch 'origin/main' into fix/gemini-cli-v1internal 
# Conflicts:
#	apps/aether-gateway/src/ai_serving/transport.rs
#	apps/aether-gateway/src/handlers/admin/provider/oauth/dispatch/batch/parse.rs
#	apps/aether-gateway/src/handlers/shared/catalog.rs
#	crates/aether-admin/src/provider/quota.rs
#	crates/aether-model-fetch/src/strategy.rs
#	crates/aether-provider-pool/src/lib.rs
#	crates/aether-provider-pool/src/service.rs
#	crates/aether-provider-transport/src/provider_types.rs
#	frontend/src/features/providers/components/ProviderDetailDrawer.vue
#	frontend/src/utils/__tests__/providerKeyQuota.spec.ts
#	frontend/src/utils/providerKeyQuota.ts
#	frontend/src/views/admin/PoolManagement.vue
 2026-05-22 17:13:57 +08:00
Mas0nShi e53d5f07e8 feat: support Gemini CLI v1internal quota 2026-05-21 15:38:10 +08:00
Entropy.Xu 9466d92a7a fix(provider): 接入 Windsurf 模型拉取和格式转换 2026-05-21 01:02:02 +08:00
AAEE86 2543676437 feat(model-fetch): fetch Kiro models from upstream 
- add Kiro ListAvailableModels request planning and headers
- route Kiro model refresh through upstream fetch
- normalize Kiro model payloads and default model metadata
- remove profileArn requirement from ListAvailableModels
 2026-05-08 13:24:27 +08:00
fawney19 f959f02d40 Implement transport profile routing 2026-05-05 22:21:23 +08:00
fawney19 1d62722d47 Improve Codex model fetching 2026-05-04 01:16:56 +08:00
fawney19 e3ea2d1451 feat: configure auth channel mismatch formats 2026-05-03 00:49:22 +08:00
fawney19 a16550249b fix: allow empty provider secrets to attempt requests 2026-04-29 18:55:52 +08:00
fawney19 e751289dfb Support per-format provider key auth 2026-04-29 15:46:50 +08:00
fawney19 07a319259b Normalize canonical API formats 2026-04-29 10:20:41 +08:00
fawney19 4ec591fbf2 centralize openai responses alias handling 2026-04-26 23:59:53 +08:00
fawney19 5b914aa78c migrate ai format conversion to responses adapters 2026-04-26 23:59:53 +08:00
RWDaiandfawney19 96a25d058b Fix OpenAI family local auth to use bearer (#302) 
* Fix OpenAI family local auth to use bearer

* test(gateway): fix bearer auth assertions for openai local flows

* test(usage): make local usage status wait resilient

* style(gateway): apply rustfmt to usage test helper

---------

Co-authored-by: fawney19 <[email protected]>
 2026-04-17 11:10:05 +08:00
AAEE86 c24a29fa65 fix(model-fetch): 提取获取计划请求，并使网关运行时与共享构建器保持一致 
- 将 build_execution_plan 的散参数收敛为 ModelFetchExecutionPlanRequest，消除 clippy too_many_arguments
- 在 aether-gateway 的 model_fetch runtime 中显式依赖 build_models_fetch_execution_plan
- 补充共享 models fetch plan builder 的运行时测试覆盖
 2026-04-12 11:57:06 +08:00
AAEE86 ab82841426 feat(model-fetch): 对齐 Rust 上游模型抓取行为到 Python 语义 
将 Rust 版上游可用模型抓取逻辑收敛到 Python 版行为，统一后台自动抓模
与管理员 provider-query 的模型发现路径，消除标准 /models、固定模型目录、
Antigravity、Vertex AI 等 provider 在两端实现上的分叉。

核心变更：
- 在 aether-model-fetch 中引入统一抓模策略层
- 覆盖标准 /models、Vertex API Key、Vertex Service Account、
  Antigravity fetchAvailableModels、固定模型目录五类抓模路径
- 将 provider-query 与后台自动抓模都切换到共享抓模入口，避免重复拼接
  URL、headers 和 provider 特判逻辑

标准模型抓取对齐：
- 按 Python 语义调整抓模优先级：
  openai:chat > openai:cli > openai:compact
  claude:chat > claude:cli
  gemini:chat > gemini:cli
- 从抓模候选中移除 openai:responses
- 为 openai:cli/openai:compact、claude:cli、gemini:* 补齐 Python 同款
  User-Agent / 浏览器指纹请求头
- Claude 抓模保留 after_id 分页语义
- Gemini 抓模统一为 v1beta/models?key=... 语义

provider-query 对齐：
- 返回结果改为按 model id 聚合，并合并/排序 api_formats
- 最终模型列表按 model id 排序，行为与 Python 保持一致
- 固定目录 provider（codex/kiro/claude_code/gemini_cli）不再依赖活跃
  endpoint，即使无 endpoint 也能返回预设模型目录
- Antigravity 多 key 查询改为按账户可用性 + tier 排序，首个成功结果即
  停止，并接入 provider 级缓存
- 仅配置 openai:responses 的 provider 不再被视为抓模成功路径

自动抓模对齐：
- 自动抓模成功时写入 allowed_models、upstream_models cache，并同步
  upstream_metadata
- upstream_metadata 合并逻辑对齐 Python，对 quota_by_model 做模型级合并，
  并保留已有 reset_time
- 自动抓模失败时不覆盖已有 allowed_models
- 固定目录 provider 在无 endpoint 场景下也可成功更新 allowed_models

Antigravity 对齐：
- 使用 POST /v1internal:fetchAvailableModels 抓取可用模型
- 按 Python 规则处理 URL fallback 和 429/404/408/5xx fallback 状态
- 强制要求 auth_config.project_id
- 过滤 Python 黑名单模型
- 解析并持久化 upstream_metadata.antigravity.quota_by_model

Vertex AI 对齐：
- API Key 模式仅抓取 publishers/google/models
- Service Account 模式新增 JWT token exchange，并按 Python region 顺序
  抓取 google + anthropic publishers
- 模型 owned_by / display_name / api_format 推断与 Python 对齐
- 软 404 处理行为与 Python 收敛

Gemini CLI / 固定目录对齐：
- Gemini CLI 改为返回 Python 预设模型目录
- 在可用时通过 loadCodeAssist 补充 plan_type/project_id 元数据
- Codex/Kiro/Claude Code 改为共享固定模型目录实现

测试：
- 扩展 aether-model-fetch 单元测试，覆盖格式优先级、openai:responses 排除、
  请求头、Claude 分页、Gemini query auth、固定目录与 metadata 合并
- 调整 provider-query 控制面测试到 Python 语义
- 新增自动抓模运行时测试，覆盖固定目录成功、Antigravity metadata 合并、
  失败保留旧 allowed_models

验证：
- cargo nextest run -p aether-model-fetch --lib
- cargo nextest run -p aether-gateway control::admin::provider_query model_fetch::runtime::tests
 2026-04-12 10:14:29 +08:00
fawney19 763ff03a7b refactor: 拆分 gateway 单体为独立 crate，新增 systemd 部署方案 
将 gateway 内部的 model-fetch、provider-transport、scheduler-core、
usage-runtime、video-tasks-core 模块提取为独立 crate；重构 gateway
内部模块结构（state/router/cache/data/query 等）；移除大量遗留模块
文件；新增 systemd 二进制部署骨架及相关文档；更新前端 usage 相关
API 和组件。
 2026-04-05 20:23:16 +08:00