Pi v0.52.7 - 更新日志

### 新功能 - 通过 `modelOverrides` 在 `models.json` 中实现按模型覆盖，允许在不替换提供商模型列表的情况下自定义内置提供商模型。详见 [docs/models.md#per-model-overrides](docs/models.md#per-model-overrides)。 - `models.json` 提供商 `models` 现在按 `id` 与内置模型合并，因此可以添加自定义模型或替换匹配的内置模型，而无需完全替换提供商。详见 [docs/models.md#overriding-built-in-providers](docs/models.md#overriding-built-in-providers)。 - 通过 `AWS_BEDROCK_SKIP_AUTH` 和 `AWS_BEDROCK_FORCE_HTTP1` 支持未经身份验证的 Bedrock 代理端点。详见 [docs/providers.md](docs/providers.md)。 ### 破坏性变更 - 将 `models.json` 提供商 `models` 的行为从完全替换更改为与内置模型按 `id` 合并。默认情况下现在保留内置模型，自定义模型按 `id` 进行 upsert（插入或更新）。 ### 新增 - 在 `models.json` 中添加 `modelOverrides`，允许在不完全替换提供商的情况下自定义每个提供商的单个内置模型 ([#1332](https://github.com/badlogic/pi-mono/pull/1332) by [@charles-cooper](https://github.com/charles-cooper)) - 添加 `AWS_BEDROCK_SKIP_AUTH` 和 `AWS_BEDROCK_FORCE_HTTP1` 环境变量，用于连接未经身份验证的 Bedrock 代理 ([#1320](https://github.com/badlogic/pi-mono/pull/1320) by [@virtuald](https://github.com/virtuald)) ### 修复 - 修复了当助手消息不包含文本时，仅思考的助手内容与后续工具执行块之间的额外间距问题 - 修复了阈值自动压缩后，排队中的引导/跟进/自定义消息卡住的问题。当代理级队列仍包含待处理消息时，通过恢复代理循环来解决 ([#1312](https://github.com/badlogic/pi-mono/pull/1312) by [@ferologics](https://github.com/ferologics)) - 修复了 `tool_result` 扩展处理程序的行为，使其在多个处理程序之间链式传递结果补丁，而不是采用“最后处理者获胜”的逻辑 ([#1280](https://github.com/badlogic/pi-mono/issues/1280)) - 修复了损坏的身份验证锁文件被优雅处理，而不是导致身份验证存储初始化崩溃的问题 ([#1322](https://github.com/badlogic/pi-mono/issues/1322)) - 修复了 Bedrock 自适应思考处理，以支持 Claude Opus 4.6 的交错思考 beta 响应 ([#1323](https://github.com/badlogic/pi-mono/pull/1323) by [@markusylisiurunen](https://github.com/markusylisiurunen)) - 修复了 OpenAI Responses API 请求默认使用 `store: false`，以避免服务器端历史记录日志记录 ([#1308](https://github.com/badlogic/pi-mono/issues/1308)) - 修复了交互式模式的启动问题，确保在资源加载后再初始化自动补全 ([#1328](https://github.com/badlogic/pi-mono/issues/1328)) - 修复了 `modelOverrides` 对嵌套对象的合并行为，并记录了使用细节 ([#1062](https://github.com/badlogic/pi-mono/issues/1062))

### New Features - Per-model overrides in `models.json` via `modelOverrides`, allowing customization of built-in provider models without replacing provider model lists. See [docs/models.md#per-model-overrides](docs/models.md#per-model-overrides). - `models.json` provider `models` now merge with built-in models by `id`, so custom models can be added or replace matching built-ins without full provider replacement. See [docs/models.md#overriding-built-in-providers](docs/models.md#overriding-built-in-providers). - Bedrock proxy support for unauthenticated endpoints via `AWS_BEDROCK_SKIP_AUTH` and `AWS_BEDROCK_FORCE_HTTP1`. See [docs/providers.md](docs/providers.md). ### Breaking Changes - Changed `models.json` provider `models` behavior from full replacement to merge-by-id with built-in models. Built-in models are now kept by default, and custom models upsert by `id`. ### Added - Added `modelOverrides` in `models.json` to customize individual built-in models per provider without full provider replacement ([#1332](https://github.com/badlogic/pi-mono/pull/1332) by [@charles-cooper](https://github.com/charles-cooper)) - Added `AWS_BEDROCK_SKIP_AUTH` and `AWS_BEDROCK_FORCE_HTTP1` environment variables for connecting to unauthenticated Bedrock proxies ([#1320](https://github.com/badlogic/pi-mono/pull/1320) by [@virtuald](https://github.com/virtuald)) ### Fixed - Fixed extra spacing between thinking-only assistant content and subsequent tool execution blocks when assistant messages contain no text - Fixed queued steering/follow-up/custom messages remaining stuck after threshold auto-compaction by resuming the agent loop when Agent-level queues still contain pending messages ([#1312](https://github.com/badlogic/pi-mono/pull/1312) by [@ferologics](https://github.com/ferologics)) - Fixed `tool_result` extension handlers to chain result patches across handlers instead of last-handler-wins behavior ([#1280](https://github.com/badlogic/pi-mono/issues/1280)) - Fixed compromised auth lock files being handled gracefully instead of crashing auth storage initialization ([#1322](https://github.com/badlogic/pi-mono/issues/1322)) - Fixed Bedrock adaptive thinking handling for Claude Opus 4.6 with interleaved thinking beta responses ([#1323](https://github.com/badlogic/pi-mono/pull/1323) by [@markusylisiurunen](https://github.com/markusylisiurunen)) - Fixed OpenAI Responses API requests to use `store: false` by default to avoid server-side history logging ([#1308](https://github.com/badlogic/pi-mono/issues/1308)) - Fixed interactive mode startup by initializing autocomplete after resources are loaded ([#1328](https://github.com/badlogic/pi-mono/issues/1328)) - Fixed `modelOverrides` merge behavior for nested objects and documented usage details ([#1062](https://github.com/badlogic/pi-mono/issues/1062))