Kimi Code @moonshot-ai/kimi-code@0.3.0

### 次要更新 - [#76](https://github.com/MoonshotAI/kimi-code/pull/76) [`6f22ae4`](https://github.com/MoonshotAI/kimi-code/commit/6f22ae48f84a062a65dcaa9510ffe96f40ab503b) - `/logout` 命令现在会打开一个选择器，让你选择要注销哪个提供商，而不是总是注销与当前模型绑定的那个。默认高亮显示当前提供商，按回车键可保持与之前相同的行为。该命令也支持使用 `/disconnect`。 ### 补丁更新 - [#62](https://github.com/MoonshotAI/kimi-code/pull/62) [`e2b2b46`](https://github.com/MoonshotAI/kimi-code/commit/e2b2b46fc9c1d6a0ada67c590b8aa56e77c9c513) - 使 `AgentRecords` 直接持有 `Agent` 实例，并将恢复分派逻辑内联。 - [#73](https://github.com/MoonshotAI/kimi-code/pull/73) [`bddc60f`](https://github.com/MoonshotAI/kimi-code/commit/bddc60f0e9af44d326dc0759a60bce93187f8a7b) - 在流式传输或压缩上下文期间，禁止运行 `/model` 和 `/sessions` 斜杠命令。 - [#70](https://github.com/MoonshotAI/kimi-code/pull/70) [`d95b013`](https://github.com/MoonshotAI/kimi-code/commit/d95b01342a7921f0863ceb37abad7984d0245509) - 保留通过 `/connect` 配置的 OpenAI 兼容模型在目录中声明的交错推理字段。 - [#78](https://github.com/MoonshotAI/kimi-code/pull/78) [`61f7d0e`](https://github.com/MoonshotAI/kimi-code/commit/61f7d0e7a2b9933bdbe7eef9177e67e7386154a2) - 使 OpenAI 兼容的推理模型在手动编写的提供商配置中开箱即用。`openai` 提供商现在通过扫描实际通用字段集（`reasoning_content`、`reasoning_details`、`reasoning`）自动检测传入响应中的思考内容，默认将思考内容序列化回传为 `reasoning_content`，并且只要对话历史包含之前的思考内容，就会自动注入 `reasoning_effort` —— 因此 DeepSeek、Qwen、One API 及其他经网关代理的服务不再需要手动设置 `reasoning_key`。`reasoning_key` 模型别名字段仍可作为非标准网关的显式覆盖选项保留。 - [#66](https://github.com/MoonshotAI/kimi-code/pull/66) [`8ddfc04`](https://github.com/MoonshotAI/kimi-code/commit/8ddfc0433e3a3a51f326116607d28b0f409e7d93) - 修复 API 密钥输入对话框在空状态下显示为遮挡圆点的问题。 - [#72](https://github.com/MoonshotAI/kimi-code/pull/72) [`0ce0072`](https://github.com/MoonshotAI/kimi-code/commit/0ce0072cb44ea2bd3a7ca9c54d141c150f0bbb77) - 修复 `~/.agents/` 目录下的用户技能无法加载的问题。 - [#86](https://github.com/MoonshotAI/kimi-code/pull/86) [`5e354d0`](https://github.com/MoonshotAI/kimi-code/commit/5e354d0cc89816228d08c3ded17e75201fb300de) - 在 TUI 中恢复运行中子代理的实时 Token 显示。 - [#57](https://github.com/MoonshotAI/kimi-code/pull/57) [`8fb61f9`](https://github.com/MoonshotAI/kimi-code/commit/8fb61f9a3ead02bbd79f3a5ab605aba26e1cb847) - 当所有待办事项均已完成后，恢复时隐藏待办面板。 - [#83](https://github.com/MoonshotAI/kimi-code/pull/83) [`7d9216d`](https://github.com/MoonshotAI/kimi-code/commit/7d9216d5aa1e96734c46c8d5d810ec7ed27b2275) - 当工具返回格式错误或缺失的结果时，始终发出配对的工具结果，防止因缺少 `tool_call_id` 导致下一次请求失败。 - [#81](https://github.com/MoonshotAI/kimi-code/pull/81) [`1fbefc9`](https://github.com/MoonshotAI/kimi-code/commit/1fbefc99398d4a8ebebb377ff7ca2846483d1a9a) - 改进 Write 工具的用户体验。 - [#79](https://github.com/MoonshotAI/kimi-code/pull/79) [`5a90b53`](https://github.com/MoonshotAI/kimi-code/commit/5a90b53b045099ecb582a36d546e90a3978f0a75) - 修复计划模式的会话重置问题，使新会话在计划审核被拒后不再失败，并在设置错误后继续接收事件。 - [#77](https://github.com/MoonshotAI/kimi-code/pull/77) [`fe60c21`](https://github.com/MoonshotAI/kimi-code/commit/fe60c215be8979f6abc8258e5255c66dd73d5a19) - 当控制终端消失时立即退出。TUI 现在处理 `SIGHUP` / `SIGTERM` 以及 stdout/stderr 的 `EIO` / `EPIPE` / `ENOTCONN` 错误，防止在父级 shell 或终端复用器意外死亡后留下独占 CPU 核心的残留 `kimi` 进程。 - [#85](https://github.com/MoonshotAI/kimi-code/pull/85) [`2bb50a3`](https://github.com/MoonshotAI/kimi-code/commit/2bb50a38d8379e2fac57547b1a563722f713c8fd) - 避免本地上下文完成上限过小，导致在生成摘要前截断推理过程。

### Minor Changes - [#76](https://github.com/MoonshotAI/kimi-code/pull/76) [`6f22ae4`](https://github.com/MoonshotAI/kimi-code/commit/6f22ae48f84a062a65dcaa9510ffe96f40ab503b) - /logout now opens a picker so you can choose which provider to log out of, instead of always logging out the one tied to the current model. The current provider is highlighted by default, so pressing Enter matches the previous behavior. The command is also available as /disconnect. ### Patch Changes - [#62](https://github.com/MoonshotAI/kimi-code/pull/62) [`e2b2b46`](https://github.com/MoonshotAI/kimi-code/commit/e2b2b46fc9c1d6a0ada67c590b8aa56e77c9c513) - Make `AgentRecords` hold the `Agent` instance directly and inline the restore dispatch logic. - [#73](https://github.com/MoonshotAI/kimi-code/pull/73) [`bddc60f`](https://github.com/MoonshotAI/kimi-code/commit/bddc60f0e9af44d326dc0759a60bce93187f8a7b) - Prevent running the `/model` and `/sessions` slash commands while streaming or compacting context. - [#70](https://github.com/MoonshotAI/kimi-code/pull/70) [`d95b013`](https://github.com/MoonshotAI/kimi-code/commit/d95b01342a7921f0863ceb37abad7984d0245509) - Preserve catalog-declared interleaved reasoning fields for OpenAI-compatible models configured through `/connect`. - [#78](https://github.com/MoonshotAI/kimi-code/pull/78) [`61f7d0e`](https://github.com/MoonshotAI/kimi-code/commit/61f7d0e7a2b9933bdbe7eef9177e67e7386154a2) - Make OpenAI-compatible reasoner models work out of the box for hand-written provider configs. The `openai` provider now auto-detects thinking on incoming responses by scanning the de facto field set (`reasoning_content`, `reasoning_details`, `reasoning`), serializes thinking back as `reasoning_content` by default, and auto-injects `reasoning_effort` whenever the conversation history contains prior thinking — so DeepSeek, Qwen, One API and other gateway-fronted services no longer require a hand-set `reasoning_key`. The `reasoning_key` model-alias field remains available as an explicit override for non-standard gateways. - [#66](https://github.com/MoonshotAI/kimi-code/pull/66) [`8ddfc04`](https://github.com/MoonshotAI/kimi-code/commit/8ddfc0433e3a3a51f326116607d28b0f409e7d93) - Fix API key input dialog showing a masked dot in empty state. - [#72](https://github.com/MoonshotAI/kimi-code/pull/72) [`0ce0072`](https://github.com/MoonshotAI/kimi-code/commit/0ce0072cb44ea2bd3a7ca9c54d141c150f0bbb77) - Fix user skills in ~/.agents/ not being loaded. - [#86](https://github.com/MoonshotAI/kimi-code/pull/86) [`5e354d0`](https://github.com/MoonshotAI/kimi-code/commit/5e354d0cc89816228d08c3ded17e75201fb300de) - Restore real-time token display for running subagents in the TUI. - [#57](https://github.com/MoonshotAI/kimi-code/pull/57) [`8fb61f9`](https://github.com/MoonshotAI/kimi-code/commit/8fb61f9a3ead02bbd79f3a5ab605aba26e1cb847) - Hide the todo panel on resume when all todos are already completed. - [#83](https://github.com/MoonshotAI/kimi-code/pull/83) [`7d9216d`](https://github.com/MoonshotAI/kimi-code/commit/7d9216d5aa1e96734c46c8d5d810ec7ed27b2275) - Always emit a paired tool result when a tool returns a malformed or missing result, preventing the next request from failing with a missing tool_call_id error. - [#81](https://github.com/MoonshotAI/kimi-code/pull/81) [`1fbefc9`](https://github.com/MoonshotAI/kimi-code/commit/1fbefc99398d4a8ebebb377ff7ca2846483d1a9a) - Improve the Write tool UX. - [#79](https://github.com/MoonshotAI/kimi-code/pull/79) [`5a90b53`](https://github.com/MoonshotAI/kimi-code/commit/5a90b53b045099ecb582a36d546e90a3978f0a75) - Fix Plan mode session resets so new sessions no longer fail after plan review rejection and continue receiving events after setup errors. - [#77](https://github.com/MoonshotAI/kimi-code/pull/77) [`fe60c21`](https://github.com/MoonshotAI/kimi-code/commit/fe60c215be8979f6abc8258e5255c66dd73d5a19) - Exit promptly when the controlling terminal goes away. The TUI now handles `SIGHUP` / `SIGTERM` and stdout/stderr `EIO` / `EPIPE` / `ENOTCONN` errors, preventing leftover `kimi` processes that pin a CPU core after the parent shell or multiplexer dies unexpectedly. - [#85](https://github.com/MoonshotAI/kimi-code/pull/85) [`2bb50a3`](https://github.com/MoonshotAI/kimi-code/commit/2bb50a38d8379e2fac57547b1a563722f713c8fd) - Avoid overly small local completion caps that can truncate reasoning before summaries are produced.

Kimi Code @moonshot-ai/[email protected]