chore: align bridge with CommandCode 0.32.1

Author: yelixir-dev

.env.example

@@ -15,7 +15,7 @@ COMMANDCODE_API_BASE=https://api.commandcode.ai
 COMMANDCODE_DEFAULT_MODEL=deepseek/deepseek-v4-pro
 COMMANDCODE_ALLOWED_MODELS=deepseek/deepseek-v4-pro,deepseek/deepseek-v4-flash
 COMMANDCODE_ALLOW_UNKNOWN_MODELS=false
-COMMANDCODE_CLI_VERSION=0.31.2
+COMMANDCODE_CLI_VERSION=0.32.1
 COMMANDCODE_TIMEOUT_MS=300000
 COMMANDCODE_EMPTY_VISIBLE_RESPONSE_POLICY=error_on_length
 

README.ko.md

@@ -67,18 +67,18 @@ curl -fsS "$BRIDGE/v1/models" | head -c 400
 
 ## 버전
 
-현재 bridge version: **v0.31.2**.
+현재 bridge version: **v0.32.1**.
 
 버전은 `/health` 응답과 웹 대시보드 오른쪽 위에 표시됩니다.
 
-### v0.31.2 CommandCode 호환성 업데이트
+### v0.32.1 CommandCode 호환성 업데이트
 
-이번 bridge release는 공식 `command-code` npm package `0.31.2`에 맞췄습니다.
+이번 bridge release는 공식 `command-code` npm package `0.32.1`에 맞췄습니다.
 
-- 기본 upstream `x-command-code-version` header는 `COMMANDCODE_CLI_VERSION`으로 덮어쓰지 않는 한 `0.31.2`을 보냅니다.
-- bridge package/runtime version도 `0.31.2`이라 `/health`, dashboard, npm metadata가 대상 CommandCode CLI version과 일치합니다.
-- `command-code@0.31.2` bundle을 직접 확인한 결과 bridge 핵심 API path(`/alpha/generate`, `/alpha/whoami`, `/alpha/billing/credits`, `/alpha/billing/subscriptions`, `/alpha/usage/summary`)는 기존 bridge 경로와 호환됩니다.
-- model catalog를 `0.31.2` CLI bundle 기준으로 갱신했습니다. 기존 enabled default는 보수적으로 유지하고, Qwen 3.6 Max Preview, MiniMax M3, MiniMax M2.5, Kimi K2.5, Step 3.5 Flash, Gemini 3.5 Flash, Gemini 3.1 Flash Lite, GLM-5, GPT 5.4/5.3 Codex/5.4 Mini, Claude Opus 4.8, 구 Claude variant 등 추가 발견 모델은 operator가 켤 때까지 disabled 상태로 포함했습니다.
+- 기본 upstream `x-command-code-version` header는 `COMMANDCODE_CLI_VERSION`으로 덮어쓰지 않는 한 `0.32.1`을 보냅니다.
+- bridge package/runtime version도 `0.32.1`이라 `/health`, dashboard, npm metadata가 대상 CommandCode CLI version과 일치합니다.
+- `command-code@0.32.1` bundle을 직접 확인한 결과 bridge 핵심 API path(`/alpha/generate`, `/alpha/whoami`, `/alpha/billing/credits`, `/alpha/billing/subscriptions`, `/alpha/usage/summary`)는 기존 bridge 경로와 호환됩니다.
+- model catalog를 `0.32.1` CLI bundle 기준으로 갱신했습니다. 기존 enabled default는 보수적으로 유지하고, Qwen 3.6 Max Preview, MiniMax M3, MiniMax M2.5, Kimi K2.5, Step 3.5 Flash, Gemini 3.5 Flash, Gemini 3.1 Flash Lite, GLM-5, GPT 5.4/5.3 Codex/5.4 Mini, Claude Opus 4.8, 구 Claude variant 등 추가 발견 모델은 operator가 켤 때까지 disabled 상태로 포함했습니다.
 
 ## 구조
 
@@ -322,7 +322,7 @@ http://<host-or-tailnet-ip>:9992/dashboard
 
 - **Header**
   - bridge online/offline 상태 표시.
-  - `v0.31.2` 같은 bridge version 표시.
+  - `v0.32.1` 같은 bridge version 표시.
 - **Server Bind**
   - local-only면 `127.0.0.1`.
   - LAN/Tailscale/VPN/reverse proxy 뒤에서만 `0.0.0.0`.
@@ -534,7 +534,7 @@ x-api-key: <BRIDGE_API_KEY>
 | `COMMANDCODE_DEFAULT_MODEL`                  | `deepseek/deepseek-v4-pro`   | `default`가 사용할 model.                                                                                  |
 | `COMMANDCODE_ALLOWED_MODELS`                 | Pro + Flash/catalog defaults | comma-separated allowlist.                                                                                 |
 | `COMMANDCODE_ALLOW_UNKNOWN_MODELS`           | `false`                      | 임의 model ID를 upstream으로 통과. 운영 비권장.                                                            |
-| `COMMANDCODE_CLI_VERSION`                    | `0.31.2`                     | upstream으로 보내는 version header.                                                                        |
+| `COMMANDCODE_CLI_VERSION`                    | `0.32.1`                     | upstream으로 보내는 version header.                                                                        |
 | `COMMANDCODE_TIMEOUT_MS`                     | `300000`                     | upstream request timeout.                                                                                  |
 | `COMMANDCODE_EMPTY_VISIBLE_RESPONSE_POLICY`  | `error_on_length`            | empty visible `finish_reason: length`를 fail-closed. `allow`는 legacy blank success 유지.                  |
 | `REQUEST_BODY_LIMIT_BYTES`                   | `1048576`                    | Fastify body limit.                                                                                        |

README.md

@@ -67,18 +67,18 @@ A healthy bridge returns JSON health/version data and a non-empty OpenAI-compati
 
 ## Version
 
-Current bridge version: **v0.31.2**.
+Current bridge version: **v0.32.1**.
 
 The version is also returned from `/health` and shown in the top-right of the web dashboard.
 
-### v0.31.2 CommandCode compatibility update
+### v0.32.1 CommandCode compatibility update
 
-This bridge release is aligned with the official `command-code` npm package `0.31.2`:
+This bridge release is aligned with the official `command-code` npm package `0.32.1`:
 
-- The default upstream `x-command-code-version` header now advertises `0.31.2` unless `COMMANDCODE_CLI_VERSION` overrides it.
-- The bridge package/runtime version is also `0.31.2` so `/health`, the dashboard, and the npm metadata match the CommandCode CLI version being targeted.
-- Direct inspection of the `command-code@0.31.2` bundle confirmed that the bridge-critical API paths remain compatible: `/alpha/generate`, `/alpha/whoami`, `/alpha/billing/credits`, `/alpha/billing/subscriptions`, and `/alpha/usage/summary`.
-- The model catalog was refreshed from the `0.31.2` CLI bundle. Existing enabled defaults stay conservative, while additional discovered entries such as Qwen 3.6 Max Preview, MiniMax M3, MiniMax M2.5, Kimi K2.5, Step 3.5 Flash, Gemini 3.5 Flash, Gemini 3.1 Flash Lite, GLM-5, GPT 5.4/5.3 Codex/5.4 Mini, Claude Opus 4.8, and older Claude variants are present but disabled by default until an operator enables them.
+- The default upstream `x-command-code-version` header now advertises `0.32.1` unless `COMMANDCODE_CLI_VERSION` overrides it.
+- The bridge package/runtime version is also `0.32.1` so `/health`, the dashboard, and the npm metadata match the CommandCode CLI version being targeted.
+- Direct inspection of the `command-code@0.32.1` bundle confirmed that the bridge-critical API paths remain compatible: `/alpha/generate`, `/alpha/whoami`, `/alpha/billing/credits`, `/alpha/billing/subscriptions`, and `/alpha/usage/summary`.
+- The model catalog was refreshed from the `0.32.1` CLI bundle. Existing enabled defaults stay conservative, while additional discovered entries such as Qwen 3.6 Max Preview, MiniMax M3, MiniMax M2.5, Kimi K2.5, Step 3.5 Flash, Gemini 3.5 Flash, Gemini 3.1 Flash Lite, GLM-5, GPT 5.4/5.3 Codex/5.4 Mini, Claude Opus 4.8, and older Claude variants are present but disabled by default until an operator enables them.
 
 ## Architecture
 
@@ -322,7 +322,7 @@ The dashboard is intentionally mobile-first. It is useful from a phone on the sa
 
 - **Header**
   - Shows bridge online/offline state.
-  - Shows bridge version, for example `v0.31.2`.
+  - Shows bridge version, for example `v0.32.1`.
 - **Server Bind**
   - Choose `127.0.0.1` for local-only use.
   - Choose `0.0.0.0` only for LAN/Tailscale/VPN/reverse-proxy use.
@@ -534,7 +534,7 @@ x-api-key: <BRIDGE_API_KEY>
 | `COMMANDCODE_DEFAULT_MODEL`                  | `deepseek/deepseek-v4-pro`   | Model used for `default`.                                                                                                    |
 | `COMMANDCODE_ALLOWED_MODELS`                 | Pro + Flash/catalog defaults | Comma-separated allowlist.                                                                                                   |
 | `COMMANDCODE_ALLOW_UNKNOWN_MODELS`           | `false`                      | Pass arbitrary model IDs upstream. Not recommended.                                                                          |
-| `COMMANDCODE_CLI_VERSION`                    | `0.31.2`                     | Version header sent upstream.                                                                                                |
+| `COMMANDCODE_CLI_VERSION`                    | `0.32.1`                     | Version header sent upstream.                                                                                                |
 | `COMMANDCODE_TIMEOUT_MS`                     | `300000`                     | Upstream request timeout.                                                                                                    |
 | `COMMANDCODE_EMPTY_VISIBLE_RESPONSE_POLICY`  | `error_on_length`            | `error_on_length` fails closed on empty visible `finish_reason: length`; `allow` preserves legacy blank success behavior.    |
 | `REQUEST_BODY_LIMIT_BYTES`                   | `1048576`                    | Fastify body limit.                                                                                                          |

README.zh.md

@@ -67,18 +67,18 @@ curl -fsS "$BRIDGE/v1/models" | head -c 400
 
 ## 版本
 
-当前 bridge version：**v0.31.2**。
+当前 bridge version：**v0.32.1**。
 
 版本也会从 `/health` 返回，并显示在 web dashboard 右上角。
 
-### v0.31.2 CommandCode 兼容性更新
+### v0.32.1 CommandCode 兼容性更新
 
-此 bridge release 已对齐官方 `command-code` npm package `0.31.2`：
+此 bridge release 已对齐官方 `command-code` npm package `0.32.1`：
 
-- 默认上游 `x-command-code-version` header 现在发送 `0.31.2`，除非用 `COMMANDCODE_CLI_VERSION` 覆盖。
-- bridge package/runtime version 也是 `0.31.2`，因此 `/health`、dashboard 和 npm metadata 与目标 CommandCode CLI version 对齐。
-- 直接检查 `command-code@0.31.2` bundle 后确认，bridge-critical API paths 仍兼容：`/alpha/generate`、`/alpha/whoami`、`/alpha/billing/credits`、`/alpha/billing/subscriptions`、`/alpha/usage/summary`。
-- model catalog 已从 `0.31.2` CLI bundle 刷新。现有 enabled defaults 保持保守；额外发现的 Qwen 3.6 Max Preview、MiniMax M3、MiniMax M2.5、Kimi K2.5、Step 3.5 Flash、Gemini 3.5 Flash、Gemini 3.1 Flash Lite、GLM-5、GPT 5.4/5.3 Codex/5.4 Mini、Claude Opus 4.8 和旧 Claude variants 等模型会保留为 disabled，直到 operator 主动启用。
+- 默认上游 `x-command-code-version` header 现在发送 `0.32.1`，除非用 `COMMANDCODE_CLI_VERSION` 覆盖。
+- bridge package/runtime version 也是 `0.32.1`，因此 `/health`、dashboard 和 npm metadata 与目标 CommandCode CLI version 对齐。
+- 直接检查 `command-code@0.32.1` bundle 后确认，bridge-critical API paths 仍兼容：`/alpha/generate`、`/alpha/whoami`、`/alpha/billing/credits`、`/alpha/billing/subscriptions`、`/alpha/usage/summary`。
+- model catalog 已从 `0.32.1` CLI bundle 刷新。现有 enabled defaults 保持保守；额外发现的 Qwen 3.6 Max Preview、MiniMax M3、MiniMax M2.5、Kimi K2.5、Step 3.5 Flash、Gemini 3.5 Flash、Gemini 3.1 Flash Lite、GLM-5、GPT 5.4/5.3 Codex/5.4 Mini、Claude Opus 4.8 和旧 Claude variants 等模型会保留为 disabled，直到 operator 主动启用。
 
 ## 架构
 
@@ -206,7 +206,7 @@ dashboard 有意采用 mobile-first 设计，适合在同一可信 tailnet 上
 
 - **Header**
   - 显示 bridge online/offline state。
-  - 显示 bridge version，例如 `v0.31.2`。
+  - 显示 bridge version，例如 `v0.32.1`。
   - 标题旁提供 Korean/English/Chinese flag language selector。
 - **Server Bind**
   - 本机专用选择 `127.0.0.1`。
@@ -373,7 +373,7 @@ x-api-key: <BRIDGE_API_KEY>
 | `COMMANDCODE_MAX_IN_FLIGHT_PER_CREDENTIAL`  | `4`                          | Per-key concurrency cap。                                                                                        |
 | `COMMANDCODE_API_BASE`                      | `https://api.commandcode.ai` | Upstream API base。除非测试明确的 alternate upstream，否则不要更改。                                             |
 | `COMMANDCODE_DEFAULT_MODEL`                 | `deepseek/deepseek-v4-pro`   | `default` 使用的 model。                                                                                         |
-| `COMMANDCODE_CLI_VERSION`                   | `0.31.2`                     | 发送给 upstream 的 version header。                                                                              |
+| `COMMANDCODE_CLI_VERSION`                   | `0.32.1`                     | 发送给 upstream 的 version header。                                                                              |
 | `COMMANDCODE_EMPTY_VISIBLE_RESPONSE_POLICY` | `error_on_length`            | `error_on_length` 对空 visible `finish_reason: length` fail closed；`allow` 保留 legacy blank success behavior。 |
 | `LOG_LEVEL`                                 | `info`                       | Fastify/Pino log level。                                                                                         |
 | `INCLUDE_REASONING`                         | `false`                      | 是否把 reasoning deltas 附加到 visible output。常规 clients 保持 false。                                         |

docs/DEPLOYMENT.ko.md

@@ -398,7 +398,7 @@ npm run smoke
 | `COMMANDCODE_DEFAULT_MODEL`        | `deepseek/deepseek-v4-pro`   | `model: "default"` 요청이 실제로 사용할 upstream model입니다.                                                            |
 | `COMMANDCODE_ALLOWED_MODELS`       | Pro + Flash                  | 허용할 model ID 목록입니다. 이 목록 밖 요청은 unknown model 허용 옵션을 켜지 않는 한 거부됩니다.                         |
 | `COMMANDCODE_ALLOW_UNKNOWN_MODELS` | `false`                      | 임의 model ID를 upstream으로 통과시킵니다. 운영에서는 권장하지 않습니다.                                                 |
-| `COMMANDCODE_CLI_VERSION`          | `0.28.1`                     | 테스트된 CommandCode CLI 동작과 맞추기 위해 upstream에 보내는 version header입니다.                                      |
+| `COMMANDCODE_CLI_VERSION`          | `0.32.1`                     | 테스트된 CommandCode CLI 동작과 맞추기 위해 upstream에 보내는 version header입니다.                                      |
 | `COMMANDCODE_TIMEOUT_MS`           | `300000`                     | upstream generation timeout입니다.                                                                                       |
 
 Credential JSON 파일 예시:

docs/DEPLOYMENT.md

@@ -399,7 +399,7 @@ npm run smoke
 | `COMMANDCODE_DEFAULT_MODEL`        | `deepseek/deepseek-v4-pro`   | Upstream model used by `model: "default"`.                                                                               |
 | `COMMANDCODE_ALLOWED_MODELS`       | Pro + Flash                  | Comma-separated allowlist. Requests outside this list are rejected unless unknown models are allowed.                    |
 | `COMMANDCODE_ALLOW_UNKNOWN_MODELS` | `false`                      | Allows arbitrary model IDs to pass through. Not recommended for production.                                              |
-| `COMMANDCODE_CLI_VERSION`          | `0.28.1`                     | Version header sent upstream to match the tested CommandCode CLI behavior.                                               |
+| `COMMANDCODE_CLI_VERSION`          | `0.32.1`                     | Version header sent upstream to match the tested CommandCode CLI behavior.                                               |
 | `COMMANDCODE_TIMEOUT_MS`           | `300000`                     | Upstream generation timeout.                                                                                             |
 
 Credential file shape:

docs/KNOW_HOW.md

@@ -54,7 +54,7 @@ Smoke-test modes:
 
 ## Billing and Multi-Key Routing
 
-CommandCode CLI 0.28.1 `/usage` uses these account endpoints with the same bearer token:
+CommandCode CLI 0.32.1 `/usage` uses these account endpoints with the same bearer token:
 
 - `GET /alpha/whoami` to discover the organization id.
 - `GET /alpha/billing/credits?orgId=...`

docs/PRD.md

@@ -54,7 +54,7 @@ Expose the local CommandCode account's DeepSeek V4 Pro route as a clean, OpenAI-
 
 ## Source Evidence
 
-- Local CommandCode 0.28.1 bundle calls `https://api.commandcode.ai/alpha/generate` and billing/usage endpoints under `/alpha/billing/*` and `/alpha/usage/summary`.
+- Local CommandCode 0.32.1 bundle calls `https://api.commandcode.ai/alpha/generate` and billing/usage endpoints under `/alpha/billing/*` and `/alpha/usage/summary`.
 - Direct Node fetch to `/alpha/whoami` and `/alpha/generate` succeeded.
 - `params.stream=false` is rejected by CommandCode; bridge must always call upstream streaming.
 - Community `pi-commandcode-provider` independently uses the same API path.

install.sh

@@ -382,7 +382,7 @@ fi
   write_env_line COMMANDCODE_DEFAULT_MODEL deepseek/deepseek-v4-pro
   write_env_line COMMANDCODE_ALLOWED_MODELS "$DEFAULT_ALLOWED_MODELS"
   write_env_line COMMANDCODE_ALLOW_UNKNOWN_MODELS false
-  write_env_line COMMANDCODE_CLI_VERSION 0.31.0
+  write_env_line COMMANDCODE_CLI_VERSION 0.32.1
   write_env_line COMMANDCODE_TIMEOUT_MS 300000
   write_env_line COMMANDCODE_EMPTY_VISIBLE_RESPONSE_POLICY error_on_length
   write_env_line COMMANDCODE_BALANCE_ALERT_ENABLED false