What's Changed
- docs(changelog): sync 2.1.1 from apps/kimi-code/CHANGELOG.md by @liruifengv in https://github.com/MoonshotAI/kimi-code/pull/4018
- chore: add regression and user-impact review to the PR workflow by @RealKai42 in https://github.com/MoonshotAI/kimi-code/pull/4023
- fix(tower): steer into notification seeded turns by @tpoisonooo in https://github.com/MoonshotAI/kimi-code/pull/3998
- fix(agent-core-v2): stop NotifyUser nudges in hosts without the update panel by @RealKai42 in https://github.com/MoonshotAI/kimi-code/pull/4054
- feat: trust workspace via KIMI_CODE_TRUST_WORKSPACE env var by @7Sageer in https://github.com/MoonshotAI/kimi-code/pull/4059
- fix(agent-core-v2): publish permission mode changes on agent.status.updated by @huangzheng2016 in https://github.com/MoonshotAI/kimi-code/pull/4057
- fix(agent-core-v2): remove interruption reminder on undo of its turn by @7Sageer in https://github.com/MoonshotAI/kimi-code/pull/4076
- feat(kimi-code): summarize workspace trust with configuration sources by @7Sageer in https://github.com/MoonshotAI/kimi-code/pull/4056
- fix(agent-core-v2): fold UserPromptSubmit hook output into the prompt message by @sailist in https://github.com/MoonshotAI/kimi-code/pull/4081
- feat: cap WaitFor at 90s and let new input end the wait by @chengluyu in https://github.com/MoonshotAI/kimi-code/pull/4061
- fix(agent-core-v2): clear inherited cron tasks at the fork record by @sailist in https://github.com/MoonshotAI/kimi-code/pull/4083
- feat(tui): add sticky user messages in fullscreen mode by @Grapedge in https://github.com/MoonshotAI/kimi-code/pull/4031
- fix(agent-core-v2): omit completion token cap unless explicitly configured by @sailist in https://github.com/MoonshotAI/kimi-code/pull/4091
- fix(tui): sanitize foreground Bash tool output by @duyphan154 in https://github.com/MoonshotAI/kimi-code/pull/2919
- feat(kimi-code): support per-model survey popup overrides by @Grapedge in https://github.com/MoonshotAI/kimi-code/pull/4033
- feat(kap-server): add allow_ignored_globs to fs:list for un-ignoring named paths by @liruifengv in https://github.com/MoonshotAI/kimi-code/pull/4058
- feat(cli): launch desktop with kimi app by @wbxl2000 in https://github.com/MoonshotAI/kimi-code/pull/4145
- fix(agent-core-v2): allow mid-turn progress notes and calm NotifyUser nudges by @RealKai42 in https://github.com/MoonshotAI/kimi-code/pull/4153
- fix(agent-core-v2): restore media tools after session resume by @7Sageer in https://github.com/MoonshotAI/kimi-code/pull/4154
- fix: report cumulative survey user turns by @Grapedge in https://github.com/MoonshotAI/kimi-code/pull/4156
- fix(tui): show Working… in the status spinner while thinking by @kimi-agent-bot in https://github.com/MoonshotAI/kimi-code/pull/4160
- fix(agent-core-v2): keep fork subagent request prefix identical to parent by @7Sageer in https://github.com/MoonshotAI/kimi-code/pull/4152
- fix: show shutdown progress and force exit on Ctrl+C during shutdown hooks by @huangzheng2016 in https://github.com/MoonshotAI/kimi-code/pull/4164
- refactor(tower): coordinate missions with deterministic orchestration by @tpoisonooo in https://github.com/MoonshotAI/kimi-code/pull/4074
- feat(agent-core-v2): add KIMI_CODE_LLM_HEADERS_TIMEOUT_MS for llm headers timeout by @sailist in https://github.com/MoonshotAI/kimi-code/pull/4092
- fix(remote-control): exit the web server when the relay takes the device offline by @sailist in https://github.com/MoonshotAI/kimi-code/pull/4182
- feat: add NotifyUser progress display to desktop and web by @RealKai42 in https://github.com/MoonshotAI/kimi-code/pull/4159
- fix: cancel in-flight video uploads when media reads are interrupted by @RealKai42 in https://github.com/MoonshotAI/kimi-code/pull/4183
- chore: remove unused kaos package and MCP kaos executor option by @7Sageer in https://github.com/MoonshotAI/kimi-code/pull/4184
- feat(agent-core-v2): report fork request prefix parity in the prompt cache probe by @7Sageer in https://github.com/MoonshotAI/kimi-code/pull/4179
- feat(tui): unify skill prompts and steering by @Grapedge in https://github.com/MoonshotAI/kimi-code/pull/4165
- feat(kap-server): add remote control device management by @liruifengv in https://github.com/MoonshotAI/kimi-code/pull/4169
- feat(agent-core-v2): distinguish session-unavailable tools and explain select_tools failures by @starquakee in https://github.com/MoonshotAI/kimi-code/pull/4190
- feat(agent-core-v2): register plugin skills under qualified names and relax plugin name validation by @oocz in https://github.com/MoonshotAI/kimi-code/pull/4188
- feat(kap-server): add remote control device deletion by @liruifengv in https://github.com/MoonshotAI/kimi-code/pull/4192
- fix(kimi-code): keep the previous turn's answer visible when a task-notification turn ends by @kimi-agent-bot in https://github.com/MoonshotAI/kimi-code/pull/3799
- feat: support model migration metadata by @7Sageer in https://github.com/MoonshotAI/kimi-code/pull/4171
- perf(tui): stop blocking the footer on git status by @Grapedge in https://github.com/MoonshotAI/kimi-code/pull/4201
- feat(agent-core-v2): add global default for deferred MCP tool disclosure by @sailist in https://github.com/MoonshotAI/kimi-code/pull/4210
- ci: publish core releases from isolated release branches by @wbxl2000 in https://github.com/MoonshotAI/kimi-code/pull/4217
- chore: sync release 2.2.0 back to main by @kimi-agent-bot in https://github.com/MoonshotAI/kimi-code/pull/4222
- docs(changelog): sync 2.2.0 from apps/kimi-code/CHANGELOG.md by @kimi-agent-bot in https://github.com/MoonshotAI/kimi-code/pull/4223
- fix(google-genai): use thinking level for Gemini 4 models by @SeleneXX in https://github.com/MoonshotAI/kimi-code/pull/4208
- fix(google-genai): map thinking effort off to LOW for Gemini 4 models by @kimi-agent-bot in https://github.com/MoonshotAI/kimi-code/pull/4230