- Store runtime selection as { provider?, model? } layers and resolve them per
field (conversation -> project -> global); drop the legacy auto provider.
- Migrate stored selections (database v49) and clear deleted profile ids.
- Re-resolve selections in Main per request and at queue dispatch.
- Add ProjectRuntimeSelector and a two-column RuntimeModelPicker; manage
link opens model connections; restore per-runtime default model tabs.
- Stop duplicating trusted instructions and history across OpenCode,
Continue and DeepSeek Harness runtimes.
- Document the layered design, global provider removal plan and planned
purpose-specific models.
Resolve 15 shared-control, overlay and responsive-layout issues plus the MCP editor being hidden behind settings. Restore theme contrast, danger and disabled states, visible keyboard focus, and consistent Continue fields. Keep the workbar add button reachable and model protocol values readable in narrow containers.
Make nested MCP and image dialogs usable above their parents with focus restoration, keep project Escape handling within the active modal, release native browser views for overlapping help, and restore keyboard access and Escape in sandboxed HTML previews without relaxing sandbox permissions. Add unit and real Electron regression coverage and update design and validation documentation.
Release note: 修复深浅主题下控件边框、危险操作、禁用状态和键盘焦点显示不一致的问题。MCP 编辑器与图片预览可在父弹窗上正常操作,HTML 全屏预览支持键盘进入和 Escape 返回;窄工作栏的添加按钮保持可用,模型连接协议文字完整可读。
Validation: the full suite recorded 5309 passed, 3 failed and 83 skipped. The ProjectSwitcher initial-focus regression was then fixed, and App, ProjectSwitcher and overlay Electron tests passed all 331 cases. The other two HeartbeatCenter failures were reproduced on isolated HEAD 11facf0 and remain pre-existing. Typecheck and lint passed after the fix; staged whitespace checks passed. The full suite was not rerun after the focused fix.
Limits: Electron evidence is from Windows development runs, with fixtures for unrelated services and native occlusion throttling disabled in some drivers. No production build, packaging, macOS/Linux, remote Agent host or external MCP connection validation; skipped tests and the documented scaling and native-select coverage gaps remain.
Model settings can now duplicate saved connections with encrypted API-key copying, while settings navigation stays usable across long forms and preserves the separate Platform Features tabs. The assistant storage startup worker now gates historical migration and compaction on actual pending legacy data, so ordinary chat free pages do not reopen the migration screen. Desktop and Agent release metadata is prepared for 0.13.8 and 0.13.2.
Release note: 改进模型连接复制、长表单设置导航和历史数据迁移判断;普通聊天退出并重开后不再重复显示迁移页面。
Projects with a saved fixed Runtime model could incorrectly claim to follow
the global Runtime configuration. Restore the fixed-model description while
keeping the simplified menu and the ability to return to inherited settings.
Model resolution, permissions, and remote Agent behavior are unchanged.
Validation: 3911 tests passed, 66 skipped; typecheck, lint, and release-note
validation passed. Added bilingual regressions for all three Agent Runtimes.
Release note: 项目已固定 Runtime 模型时,现在明确显示所用固定模型,不再误报为跟随全局设置。
Restore Ask workspace reads and output paging without granting write access,
and let programming Subagents reuse their parent's browser page. Allocate
browser resources only when used, keep Agent pages aligned with workbar tabs,
recover released bindings, and remove unused automatic screenshot work.
Preserve inherited Runtime selections during remote usage persistence and
manual compression. Keep viable channel model fallbacks, display the actual
scheduled-task mode and Harness model source, and avoid rediscovery delays
when calling healthy MCP tools. Fix directory/history refresh, case-only
renames, rename-aware commit diffs, and FastGPT probe success reporting.
Add regression coverage and align feature documentation. The Agent-side
connection timeout fix requires deploying an updated Agent; existing Ask
read-only and Execute authorization boundaries remain unchanged.
Final validation: 3839 tests passed, 66 skipped; typecheck, lint, development
build, and diff checks passed. Isolated Electron UI/MCP workflows passed with
a loopback model fixture. Current-source Linux x64 Host checks passed with
three successful real model requests. Chromium loadURL rejection was not
reproduced in the UI; released-binding recovery has automated coverage.
Release note: 修复 Ask 读取项目文件、浏览器页面显示与恢复、文件管理及模型配置跟随中的不一致,并减少无用截图和 MCP 等待。定时任务显示实际工作模式,远程会话可正常保存继承 Runtime 的用量;远端连接超时修复需更新 Agent 后生效。
Model connections can now save bounded JSON request headers and top-level body fields. Direct requests and Continue support both; local OpenCode and DeepSeek Harness use only native header support, while the managed remote OpenCode gateway applies both.
Runtime, protocol, and authentication fields keep precedence. Custom values are ordinary connection settings, not credential storage. This migrates Runtime settings from v20 to v21 and updates the DeepSeek control and managed Runtime ACP capabilities.
Release note: 模型连接详情新增自定义请求头和请求体;直连与 Continue 支持两者,本地 OpenCode 和 DeepSeek Harness 仅应用原生支持的请求头,受管远程 OpenCode 支持两者,界面会明确提示生效范围。