feat(cursor): add HTTP/1.1 compatibility transport

docs-site/src/content/docs/guides/providers.md

@@ -116,7 +116,7 @@ ocx logout <provider>
 | `nous` | `openai-chat` | `https://inference-api.nousresearch.com/v1` | Nous Research subscription gateway (same backend Hermes Agent uses). Device-grant login against `portal.nousresearch.com`; the access token is the per-request inference JWT. Mixed paid + `:free` model catalog (`tencent/hy3:free`, `stepfun/step-3.7-flash:free`, ...) discovered live from the signed-in account. Refresh tokens are single-use and rotated on every refresh. |
 | `kiro` | `kiro` | `https://runtime.us-east-1.kiro.dev` | Initial login imports the installed, signed-in `kiro-cli` session (on Unix, install with `curl -fsSL https://cli.kiro.dev/install` &#124; `bash`; on Windows PowerShell, use `irm 'https://cli.kiro.dev/install.ps1'` &#124; `iex`; then run `kiro-cli login`). **Add account** logs `kiro-cli` out, starts a fresh browser login that switches the account used by `kiro-cli`, and stores account-scoped profile metadata. Existing OpenCodex accounts are preserved, and cancellation or failure restores the previous `kiro-cli` session. |
 | `google-antigravity` | `google` | `https://daily-cloudcode-pa.googleapis.com` | Google OAuth over the Cloud Code Assist wire. Live discovery uses CCA's authenticated `v1internal:fetchAvailableModels` endpoint and publishes the agent models available to the signed-in account; the maintained catalog remains the fallback. |
-| `cursor` | `cursor` | `https://api2.cursor.sh` | Experimental PKCE login, live HTTP/2 transport, and account-filtered model discovery. |
+| `cursor` | `cursor` | `https://api2.cursor.sh` | Experimental PKCE login, live HTTP/2 transport with an opt-in HTTP/1.1 compatibility path, and account-filtered model discovery. |
 | `github-copilot` | `openai-chat` | `https://api.githubcopilot.com` | Experimental. GitHub device flow + `copilot_internal` exchange (VS Code OAuth client). Requires an active Copilot subscription; not an official third-party API. |
 
 After a terminal Nous refresh failure, run `ocx login nous` to reauthenticate.
@@ -511,7 +511,9 @@ provider-wide adapter. To opt a model without a built-in default (for example
 Cursor is tracked separately as an experimental adapter. `adapter: "cursor"` appears in `ocx init`
 and the dashboard Add Provider picker as an experimental local config entry with Cursor's static
 fallback model catalog metadata. When a Cursor access token is configured, opencodex uses Cursor's
-live HTTP/2 transport. Its bundled fallback seed includes `gpt-5.6-sol` / `terra` / `luna` (1M context),
+live HTTP/2 transport. Set `upstreamHttpVersion: "http1.1"` when a proxy requires Cursor's HTTP/1.1
+compatibility path; the setting covers both inference and live model discovery and is exposed at
+**Providers → Cursor → Settings → Cursor transport**. Its bundled fallback seed includes `gpt-5.6-sol` / `terra` / `luna` (1M context),
 regular/Fast rows for Grok 4.5 and 4.6 (500K), and `kimi-k3` (262K); live discovery decides which
 remain visible for the account. Grok 4.6 exposes `low` / `medium` / `high` / `xhigh` in both forms,
 while 4.5 stops at `high`. Fast requests send the matching base Grok model with separate `effort`

docs-site/src/content/docs/reference/adapters.md

@@ -186,7 +186,10 @@ advertised effort control on those models as proof of upstream-native reasoning
 
 ## `cursor`
 
-**Targets:** Cursor's `agent.v1.AgentService/Run` over HTTP/2 Connect streaming at `api2.cursor.sh`.
+**Targets:** Cursor's `agent.v1.AgentService/Run` over HTTP/2 Connect streaming at `api2.cursor.sh`
+by default. With `upstreamHttpVersion: "http1.1"` (or `"h1"`), uses Cursor's HTTP/1.1
+compatibility pair: `agent.v1.AgentService/RunSSE` for server output and
+`aiserver.v1.BidiService/BidiAppend` for client messages.
 **Auth:** Cursor OAuth/access token from `provider.apiKey` or the forwarded authorization header.
 
 - Uses `runTurn` rather than the ordinary fetch/parse path. Requests, server events, tool arguments,
@@ -195,6 +198,8 @@ advertised effort control on those models as proof of upstream-native reasoning
 - Replays conversation state through content-addressed blobs, maps server tool calls back to Codex,
   discovers live Cursor models through the protobuf `GetUsableModels` RPC, and retries only before a
   run request is committed to the wire.
+- Honors `upstreamHttpVersion` for both live model discovery and inference. `auto`, `http2`, and `h2`
+  preserve the existing HTTP/2 transport; only `http1.1` and `h1` select compatibility mode.
 - Exposes Cursor Router as `cursor/auto` plus explicit `cursor/auto-cost`,
   `cursor/auto-balance`, and `cursor/auto-intelligence` entries. Explicit levels are encoded in
   `requested_model.parameters` while the legacy `cursor/auto` entry retains the account/team default.

docs-site/src/content/docs/reference/configuration/providers.md

@@ -67,7 +67,7 @@ differing backup and rewrites known legacy namespaced selected ids to bare ids.
 | `adapter` | `string` | One of `openai-chat`, `openai-responses`, `anthropic`, `google`, `kiro`, `cursor`, `azure-openai` (or alias `azure`). |
 | `baseUrl` | `string` | Upstream API base URL. Most built-in fixed endpoints ignore a mismatch; collision-safe key presets preserve an older same-named custom destination. |
 | `requestPacing?` | `{ enabled, requestsPerMinute?, minIntervalMs?, models? }` | Optional client-side outbound request-start pacing, separate from upstream usage, billing, and rate-limit indicators. RPM is converted to an even interval; `minIntervalMs` may impose a longer interval. Provider limits apply across all models, while `models` entries use exact upstream model IDs (for example `nvidia/llama-3.1-nemotron-ultra-253b-v1`) and can only add delay. Queue waits do not consume the upstream response-header timeout. HTTP, Responses WebSocket, and explicit adapter `fetchResponse`/`runTurn` dispatches are covered. |
-| `upstreamHttpVersion?` | `"auto" \| "http1.1" \| "h1" \| "http2" \| "h2"` | Pin the HTTP version used for upstream requests to this provider. Defaults to `auto`, which lets Bun negotiate. Set `http1.1` when a provider's HTTP/2 SSE stream stalls instead of delivering events — the symptom is a long-running streaming request that produces nothing and eventually times out. Management `POST`/`PATCH` accept `null` to clear it back to `auto`. |
+| `upstreamHttpVersion?` | `"auto" \| "http1.1" \| "h1" \| "http2" \| "h2"` | Pin the HTTP version used for upstream requests to this provider. Defaults to `auto`, which lets Bun negotiate. An explicit pin requires an HTTPS target and fails locally when it cannot be honored. Set `http1.1` when a provider's HTTP/2 SSE stream stalls instead of delivering events — the symptom is a long-running streaming request that produces nothing and eventually times out. For Cursor, `http1.1`/`h1` selects its `RunSSE` + `BidiAppend` compatibility transport for inference and also pins live model discovery. Management `POST`/`PATCH` accept `null` to clear it back to `auto`. |
 | `responsesPath?` | `string` | Relative resource path for key-auth `openai-responses` requests. It must start with `/` and contain no scheme, query, or fragment. |
 | `supportsServiceTier?` | `boolean` | Tri-state canonical Fast capability fallback. `true` publishes Fast in the catalog, satisfies service-tier routing requirements, contributes a supported fingerprint, and lets fast mode inject the provider's canonical wire value on a compatible final adapter. `false` strips the field and never injects, and exact model declarations cannot reopen it. Absent leaves the provider unclassified: fast mode does not inject or normalize a canonical caller value, and caller values obey the final wire's forwarding permission (`chatServiceTier` on Chat; passthrough on Responses). The registry classifies canonical OpenAI (`true`), DeepSeek, and Volcengine Ark (`false`); set it explicitly only for custom gateways that genuinely support tiers. |
 | `modelSupportsServiceTier?` | `Record<string, boolean>` | Exact upstream model capability overrides. Exact `true` enables canonical Fast for that model; exact `false` narrows provider defaults. An explicit provider-level `supportsServiceTier: false` remains fail-closed and cannot be reopened. Exact `true` does not authorize foreign caller-tier forwarding on Chat. Undeclared models fall back to provider-wide behavior. Management `PATCH /api/providers` merges entries and accepts `null` to clear one. |
@@ -294,6 +294,14 @@ so passthrough stays byte-for-byte identical.
 ## Cursor provider (`adapter: "cursor"`)
 
 The Cursor bridge is experimental. After `ocx login cursor`, add or edit `providers.cursor`.
+
+If a proxy cannot carry Cursor's default HTTP/2 stream, set `upstreamHttpVersion` to `"http1.1"`
+or its `"h1"` alias.
+This switches inference to Cursor's `RunSSE` + `BidiAppend` compatibility transport and uses
+HTTP/1.1 for `GetUsableModels` discovery as well. The value requires an HTTPS `baseUrl`. Leave it
+unset or use `"auto"` for the existing HTTP/2 behavior. In the dashboard choose
+**Providers → Cursor → Settings → Cursor transport**.
+
 Cursor Router's optimization ladder is exposed as separate Codex ids because the picker cannot render
 Cursor-specific model parameters:
 
@@ -330,8 +338,8 @@ Cursor server-driven local tools are disabled by default. Codex continues using
 }
 ```
 
-Set the field on `providers.cursor`, not at the top level. In the dashboard use **Providers → Cursor
-→ Edit JSON**, save, then restart. Legacy `unsafeAllowNativeLocalExec: true` equals
+Set `nativeLocalExec` on `providers.cursor`, not at the top level. In the dashboard use **Providers
+→ Cursor → Edit JSON**, save, then restart. Legacy `unsafeAllowNativeLocalExec: true` equals
 `nativeLocalExec: "on"` only when `nativeLocalExec` is unset. MCP, screen recording, and computer use
 are controlled separately by `mcpServers` and `desktopExecutor`.
 

docs-site/src/content/docs/zh-cn/guides/providers.md

@@ -101,7 +101,7 @@ ocx logout <provider>
 | `nous` | `openai-chat` | `https://inference-api.nousresearch.com/v1` | Nous Research 订阅网关（与 Hermes Agent 使用同一后端）。通过设备授权登录 `portal.nousresearch.com`；access 令牌是每个请求的 inference JWT。付费 + `:free` 模型混合目录（`tencent/hy3:free`、`stepfun/step-3.7-flash:free` 等）会从已登录账户实时发现。Refresh 令牌是单次使用，每次刷新都会轮换。 |
 | `kiro` | `kiro` | `https://runtime.us-east-1.kiro.dev` | 首次登录会导入已安装并已登录的 Kiro CLI 会话（Unix 使用 `curl -fsSL https://cli.kiro.dev/install` &#124; `bash`；Windows PowerShell 使用 `irm 'https://cli.kiro.dev/install.ps1'` &#124; `iex`；然后运行 `kiro-cli login`）。**添加账户**会先退出 `kiro-cli`，再启动新的浏览器登录，从而切换 `kiro-cli` 自身使用的账户，并保存账户范围的配置文件元数据。现有 OpenCodex 账户会保留；如果取消或失败，则恢复之前的 `kiro-cli` 会话。 |
 | `google-antigravity` | `google` | `https://daily-cloudcode-pa.googleapis.com` | 通过 Cloud Code Assist 协议使用 Google OAuth。实时发现调用已认证的 CCA `v1internal:fetchAvailableModels` 端点，并仅发布当前登录账户可用的 agent 模型；维护中的目录仍作为回退。 |
-| `cursor` | `cursor` | `https://api2.cursor.sh` | 实验性 PKCE 登录、HTTP/2 传输和按账号筛选的模型发现。 |
+| `cursor` | `cursor` | `https://api2.cursor.sh` | 实验性 PKCE 登录、带可选 HTTP/1.1 兼容路径的 HTTP/2 传输，以及按账号筛选的模型发现。 |
 | `github-copilot` | `openai-chat` | `https://api.githubcopilot.com` | 实验性。GitHub 设备流 + `copilot_internal` 交换（VS Code OAuth 客户端）。需要有效的 Copilot 订阅；不是官方第三方 API。 |
 
 Nous refresh 发生终止性失败后，请运行 `ocx login nous` 重新认证。
@@ -365,7 +365,9 @@ adapter。若要将没有内置默认值的模型（例如 `gpt-5.4-nano`）接
 
 Cursor 作为单独的实验性 adapter 进行跟踪。`adapter: "cursor"` 会作为实验性本地配置出现在
 `ocx init` 和 dashboard Add Provider picker 中，并保存 Cursor 的静态回退模型目录 metadata。配置
-Cursor access token 后，opencodex 会使用 Cursor live HTTP/2 transport。内置回退列表包含上下文为
+Cursor access token 后，opencodex 会使用 Cursor live HTTP/2 transport。代理要求 Cursor 的
+HTTP/1.1 兼容路径时，可设置 `upstreamHttpVersion: "http1.1"`；该设置同时覆盖推理与实时模型发现，
+并可在 **Providers → Cursor → 设置 → Cursor 传输协议** 中选择。内置回退列表包含上下文为
 1M 的 `gpt-5.6-sol` / `terra` / `luna`、上下文为 500K 的 Grok 4.5/4.6 普通与 Fast 条目，以及上下文为
 262K 的 `kimi-k3`；最终显示哪些模型由账号的实时发现结果决定。Grok 4.6 的两种形式均提供
 `low` / `medium` / `high` / `xhigh`，而 4.5 最高为 `high`。Fast 请求会发送对应的 Grok 基础模型，

docs-site/src/content/docs/zh-cn/reference/adapters.md

@@ -123,15 +123,19 @@ Kiro 的 assistant 文本本身没有可靠的回合结束标记，但终止的
 
 ## `cursor`
 
-**目标：** `api2.cursor.sh` 上采用 HTTP/2 Connect streaming 的
-`agent.v1.AgentService/Run`。
+**目标：** 默认使用 `api2.cursor.sh` 上采用 HTTP/2 Connect streaming 的
+`agent.v1.AgentService/Run`。配置 `upstreamHttpVersion: "http1.1"`（或 `"h1"`）后，改用
+Cursor 的 HTTP/1.1 兼容传输：通过 `agent.v1.AgentService/RunSSE` 接收 server output，并通过
+`aiserver.v1.BidiService/BidiAppend` 发送 client message。
 **认证：** `provider.apiKey` 或转发 authorization header 中的 Cursor OAuth/access token。
 
 - 使用 `runTurn`，而不是常规 fetch/parse 路径。请求、server event、工具参数、usage checkpoint
   和 client reply 由 `cursor/gen/agent_pb.ts` 中的 `@bufbuild/protobuf` schema 编码，并 frame 成
   Connect message。
 - 经 content-addressed blob 重放对话状态，把 server tool call 映射回 Codex，用 protobuf
   `GetUsableModels` RPC 发现实时 Cursor 模型，并且只在 run request 尚未 commit 到 wire 前重试。
+- 模型实时发现和推理都会遵守 `upstreamHttpVersion`。`auto`、`http2` 与 `h2` 保持原有 HTTP/2
+  transport；只有 `http1.1` 与 `h1` 会选择兼容模式。
 - 保留 `cursor/grok-4.5-fast` 作为可选模型，但向 Cursor 发送规范的 `grok-4.5` 模型，并将独立的
   `effort` 和 `fast=true` 值放入 `requested_model.parameters`。
 - Cursor 原生本地 filesystem/shell/network 执行默认被拒绝。显式 `mcpServers` 与

docs-site/src/content/docs/zh-cn/reference/configuration/providers.md

@@ -220,7 +220,15 @@ affinity。这些策略不能规避 provider enforcement。
 
 ## Cursor 提供者（`adapter: "cursor"`）
 
-Cursor 桥接是实验性的。执行 `ocx login cursor` 之后，添加或编辑 `providers.cursor`。Cursor Router 的优化层级会作为独立的 Codex id 暴露，因为选择器无法渲染 Cursor 特定的模型参数：
+Cursor 桥接是实验性的。执行 `ocx login cursor` 之后，添加或编辑 `providers.cursor`。
+
+如果代理无法承载 Cursor 默认的 HTTP/2 stream，请将 `upstreamHttpVersion` 设置为
+`"http1.1"` 或其别名 `"h1"`。推理会切换到 Cursor 的 `RunSSE` + `BidiAppend` 兼容传输，`GetUsableModels`
+实时发现也会使用 HTTP/1.1。该配置要求 `baseUrl` 使用 HTTPS。保持未设置或使用 `"auto"`，
+则继续使用现有 HTTP/2 行为。
+在仪表板中，可通过 **Providers → Cursor → 设置 → Cursor 传输协议** 进行选择。
+
+Cursor Router 的优化层级会作为独立的 Codex id 暴露，因为选择器无法渲染 Cursor 特定的模型参数：
 
 | Codex model | Cursor Router mode |
 | --- | --- |
@@ -251,7 +259,7 @@ Cursor 由服务端驱动的本地工具默认是禁用的。Codex 继续使用
 }
 ```
 
-请将该字段设置在 `providers.cursor` 上，而不是顶层。在仪表板中，使用 **Providers → Cursor → Edit JSON**，保存，然后重启。旧的 `unsafeAllowNativeLocalExec: true` 仅在未设置 `nativeLocalExec` 时，才等同于 `nativeLocalExec: "on"`。MCP、屏幕录制和 computer use 由 `mcpServers` 和 `desktopExecutor` 单独控制。
+请将 `nativeLocalExec` 设置在 `providers.cursor` 上，而不是顶层。在仪表板中，使用 **Providers → Cursor → Edit JSON**，保存，然后重启。旧的 `unsafeAllowNativeLocalExec: true` 仅在未设置 `nativeLocalExec` 时，才等同于 `nativeLocalExec: "on"`。MCP、屏幕录制和 computer use 由 `mcpServers` 和 `desktopExecutor` 单独控制。
 
 每个 `mcpServers.<name>` 都可以接受 `command`（stdio）或 `url`（Streamable HTTP）。stdio 还接受 `args`、`env` 和 `cwd`；HTTP 接受 `headers`。两者都支持 `enabled`（默认 true）和 `toolPrefix`。`desktopExecutor` 接受 `computerUseCommand`、`recordScreenCommand`、`cwd`、`env` 和 `timeoutMs`（默认 `30000`）。命令通过 `sh -c` 执行，从 stdin 读取一个 JSON 请求，并且必须向 stdout 写入一个 JSON 结果。