@moonshot-ai/kimi-code@0.0.0-dev.2

What's Changed

• docs(changelog): sync 2.1.1 from apps/kimi-code/CHANGELOG.md by @liruifengv in #4018
• chore: add regression and user-impact review to the PR workflow by @RealKai42 in #4023
• fix(tower): steer into notification seeded turns by @tpoisonooo in #3998
• fix(agent-core-v2): stop NotifyUser nudges in hosts without the update panel by @RealKai42 in #4054
• feat: trust workspace via KIMI_CODE_TRUST_WORKSPACE env var by @7Sageer in #4059
• fix(agent-core-v2): publish permission mode changes on agent.status.updated by @huangzheng2016 in #4057
• fix(agent-core-v2): remove interruption reminder on undo of its turn by @7Sageer in #4076
• feat(kimi-code): summarize workspace trust with configuration sources by @7Sageer in #4056
• fix(agent-core-v2): fold UserPromptSubmit hook output into the prompt message by @sailist in #4081
• feat: cap WaitFor at 90s and let new input end the wait by @chengluyu in #4061
• fix(agent-core-v2): clear inherited cron tasks at the fork record by @sailist in #4083
• feat(tui): add sticky user messages in fullscreen mode by @Grapedge in #4031
• fix(agent-core-v2): omit completion token cap unless explicitly configured by @sailist in #4091
• fix(tui): sanitize foreground Bash tool output by @duyphan154 in #2919
• feat(kimi-code): support per-model survey popup overrides by @Grapedge in #4033
• feat(kap-server): add allow_ignored_globs to fs:list for un-ignoring named paths by @liruifengv in #4058
• feat(cli): launch desktop with kimi app by @wbxl2000 in #4145
• fix(agent-core-v2): allow mid-turn progress notes and calm NotifyUser nudges by @RealKai42 in #4153
• fix(agent-core-v2): restore media tools after session resume by @7Sageer in #4154
• fix: report cumulative survey user turns by @Grapedge in #4156
• fix(tui): show Working… in the status spinner while thinking by @kimi-agent-bot in #4160
• fix(agent-core-v2): keep fork subagent request prefix identical to parent by @7Sageer in #4152
• fix: show shutdown progress and force exit on Ctrl+C during shutdown hooks by @huangzheng2016 in #4164
• refactor(tower): coordinate missions with deterministic orchestration by @tpoisonooo in #4074
• feat(agent-core-v2): add KIMI_CODE_LLM_HEADERS_TIMEOUT_MS for llm headers timeout by @sailist in #4092
• fix(remote-control): exit the web server when the relay takes the device offline by @sailist in #4182
• feat: add NotifyUser progress display to desktop and web by @RealKai42 in #4159
• fix: cancel in-flight video uploads when media reads are interrupted by @RealKai42 in #4183
• chore: remove unused kaos package and MCP kaos executor option by @7Sageer in #4184
• feat(agent-core-v2): report fork request prefix parity in the prompt cache probe by @7Sageer in #4179
• feat(tui): unify skill prompts and steering by @Grapedge in #4165
• feat(kap-server): add remote control device management by @liruifengv in #4169
• feat(agent-core-v2): distinguish session-unavailable tools and explain select_tools failures by @starquakee in #4190
• feat(agent-core-v2): register plugin skills under qualified names and relax plugin name validation by @oocz in #4188
• feat(kap-server): add remote control device deletion by @liruifengv in #4192
• fix(kimi-code): keep the previous turn's answer visible when a task-notification turn ends by @kimi-agent-bot in #3799
• feat: support model migration metadata by @7Sageer in #4171
• perf(tui): stop blocking the footer on git status by @Grapedge in #4201
• feat(agent-core-v2): add global default for deferred MCP tool disclosure by @sailist in #4210
• ci: publish core releases from isolated release branches by @wbxl2000 in #4217
• chore: sync release 2.2.0 back to main by @kimi-agent-bot in #4222
• docs(changelog): sync 2.2.0 from apps/kimi-code/CHANGELOG.md by @kimi-agent-bot in #4223
• fix(google-genai): use thinking level for Gemini 4 models by @SeleneXX in #4208
• fix(google-genai): map thinking effort off to LOW for Gemini 4 models by @kimi-agent-bot in #4230
• docs(tower): add description by @tpoisonooo in #4216
• feat(oauth): parse unified /models schema for managed Kimi Code models by @liruifengv in #4209
• feat(tui): auto-generate session title after the first turn by @7Sageer in #4200
• fix(QuestionDialog): add back navigation when AskUserQuestion by @gaoyuan1223m in #4186
• fix(agent-core-v2): nudge the agent to keep its todo list current by @RealKai42 in #4239
• fix: align subagent tool descriptions with availability by @7Sageer in #4066
• fix(server): run agent hooks in the session's working directory by @wbxl2000 in #4242
• feat(remote-control): support chunked streaming transfers up to 100 MiB by @sailist in #4214
• feat(agent-core-v2): let dangerous_command_guard=false disable all built-in guard ask policies by @sailist in #4250
• fix(tui): fold previews only when collapsing saves rows by @kimi-agent-bot in #4248
• feat: notify the agent of external changes after /reload by @liruifengv in #4019
• docs: add review scope rules to AGENTS.md by @kimi-agent-bot in #4265
• feat: add a default workspace for chats without a project by @wbxl2000 in #4224
• feat(server): place the default workspace under the home folder by @wbxl2000 in #4277
• fix(kap-server): evict session event state on session close and archive by @sailist in #4281
• feat: share session rating rules across CLI and server by @Grapedge in #4272
• test(remote-control): split the tunnel end-to-end case and make transfer limits injectable by @kimi-agent-bot in #4290
• feat(agent-core-v2): add cross-process session occupancy lock by @7Sageer in #4246
• fix(transcript): realign rebuilt turn ordinals to wire ids and mark wedged turns by @sailist in #4280
• docs(changelog): sync 2.3.0 from apps/kimi-code/CHANGELOG.md by @kimi-agent-bot in #4305
• chore: sync release/2026-10-21 (2.3.0) back to main by @kimi-agent-bot in #4306
• perf(agent-core-v2): speed up large session resume by @huangzheng2016 in #4313
• feat(kap-server): reclaim idle sessions by @7Sageer in #4284
• fix(agent-core-v2): reclaim session locks from zombie processes by @7Sageer in #4320
• feat: add /provider refresh subcommand by @huangzheng2016 in #4316
• feat: let users take over a session lock left by an exited process by @7Sageer in #4328
• chore: upgrade to TypeScript 7 (native compiler) by @kimi-agent-bot in #4329
• feat(tui): allow expanding streaming thinking by @Grapedge in #4322
• refactor: remove unused session terminal service by @wbxl2000 in #4333
• fix(agent-core-v2): update import-from-cc-codex for current Claude Code and Codex layouts by @RealKai42 in #4331
• feat(tower): improve concurrency, locking, and per-spawn model selection by @tpoisonooo in #4312
• feat(agent-core-v2): leave Plan mode to the user and review every plan by @RealKai42 in #4247
• docs(changelog): sync 2.3.1 from apps/kimi-code/CHANGELOG.md by @kimi-agent-bot in #4360
• chore: sync release/2026-10-23 (2.3.1) back to main by @kimi-agent-bot in #4361
• feat(oauth): infer region for logged-out installs by @sailist in #4311
• test: isolate cloud config in message flow tests by @Grapedge in #4363
• fix(agent-core-v2): wait for MCP discovery before tool announcements by @pvzheroes125 in #4364
• docs(changelog): sync 2.3.2 from apps/kimi-code/CHANGELOG.md by @kimi-agent-bot in #4388
• chore: sync release/2026-10-25 (2.3.2) back to main by @kimi-agent-bot in #4387
• feat: import Claude Code and Codex sessions by @RealKai42 in #4342
• fix(agent-core-v2): truncate tool results on code point boundaries by @kimi-agent-bot in #4365
• fix(agent-core-v2): stop stripping dynamic tool context from compaction requests by @RealKai42 in #4410
• feat(tui): refine fullscreen docks and transcript cards by @Grapedge in #4310
• fix(tui): show compacting activity with working tips by @Grapedge in #4346
• fix(agent-core-v2): drop empty OpenAI tool declaration messages by @sailist in #4421
• fix(agent-core-v2): close idle MCP OAuth callback connections by @pvzheroes125 in #4431
• fix(server): allow data and blob media fetches in CSP by @kimi-agent-bot in #4434
• fix(server): allow Shiki WASM in CSP by @kimi-agent-bot in #4435
• fix(workspace): reclaim instances when last owner releases by @sailist in #4422
• feat(tui): support OSC 7501 program status by @Grapedge in #4347
• docs(changelog): sync 2.3.3 from apps/kimi-code/CHANGELOG.md by @kimi-agent-bot in #4445
• fix(agent-core-v2): return a stopped subagent's partial output to the parent agent by @RealKai42 in #4430
• fix(agent-core-v2): keep media recovery across turns and compaction by @7Sageer in #4413
• fix(agent-core-v2): reconnect MCP servers automatically after transient failures by @sailist in #4283
• chore: sync release/2026-10-26 (2.3.3) back to main by @kimi-agent-bot in #4453
• feat(kap-server): push approvals and questions as interaction frames by @wbxl2000 in #4428
• feat: add independent dev release channel by @Grapedge in #4425
• fix(agent-core-v2): scope verification to the change and stop repeating passed checks by @RealKai42 in #4451
• fix(agent-core-v2): answer questions without edits and clean up test leftovers by @RealKai42 in #4450
• fix(kap-server): apply swarm mode from prompt submissions by @kimi-agent-bot in #4460
• feat: support symlinked directories in file tree by @liruifengv in #4457
• fix(agent-core-v2): keep the prompt cache through auto compaction by @7Sageer in #4456
• refactor(kap-server): merge WebSocket session subscriptions into subscribe_v2 by @wbxl2000 in #4470
• fix(dev): avoid CI run response buffer overflow by @kimi-agent-bot in #4469

New Contributors

• @duyphan154 made their first contribution in #2919

Full Changelog: https://github.com/MoonshotAI/kimi-code/compare/@moonshot-ai/kimi-code@2.1.1...@moonshot-ai/kimi-code@0.0.0-dev.2