* feat(core): add microcompaction for idle context cleanup
Clear old tool result content from chat history when the user returns
after an idle period (default 60 min). Replaces functionResponse output
with a sentinel string for compactable tools (read_file, shell, grep,
glob, web_fetch, web_search, edit, write_file), keeping the N most
recent results intact (default 5). Runs before full compression so it
can shed tokens cheaply without an API call.
- Time-based trigger reuses lastApiCompletionTimestamp from thinking cleanup
- Per-part counting so keepRecent applies to individual tool results
even when batched in parallel
- Preserves tool error responses (only clears successful outputs)
- Configurable via settings.json (context.microcompaction) with env var
overrides for E2E testing
- Enabled by default
* refactor(config): unify idle cleanup settings under clearContextOnIdle
Consolidate thinking block cleanup and tool results microcompaction
config into a single `context.clearContextOnIdle` settings group:
{
"context": {
"clearContextOnIdle": {
"thinkingThresholdMinutes": 5,
"toolResultsThresholdMinutes": 60,
"toolResultsNumToKeep": 5
}
}
}
- Use -1 on either threshold to disable that cleanup (no enabled bool)
- Remove separate `microcompaction` and `gapThresholdMinutes` settings
- Thinking cleanup: 5 min default (unchanged)
- Tool results cleanup: 60 min default
- Preserve tool error responses (only clear successful outputs)
* feat(vscode-ide-companion): add clearContextOnIdle settings configuration
- Add gapThresholdMinutes settings for thinking blocks, tool results, and retention count
- Remove deprecated gapThresholdMinutes from root settings level
This reorganizes the context clearing settings into a dedicated clearContextOnIdle object with configurable thresholds for thinking blocks and tool results.
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
* fix(core): restrict microcompaction to user-initiated messages only
Move microcompactHistory() inside the UserQuery/Cron guard so model
latency during tool-call loops doesn't count as user idle time.
* docs: update settings docs for clearContextOnIdle config rename
Replace stale `context.gapThresholdMinutes` entry with the new
`context.clearContextOnIdle.*` settings group introduced in the
microcompaction feature.
* fix(core): address review comments on microcompaction PR
- Guard against NaN in toolResultsNumToKeep with Number.isFinite()
- Report effective keepRecent (after Math.max) in meta, not raw config
- Fix comment to mention cron messages alongside user messages
---------
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
Align with observed provider prompt-cache TTL (~5 min). Add
`context.gapThresholdMinutes` setting so users can tune the threshold
for providers with different cache TTLs.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Change cron/loop tools from opt-out to opt-in. Cron tools are now
disabled by default and can be enabled via:
- settings.json: { "experimental": { "cron": true } }
- Environment variable: QWEN_CODE_ENABLE_CRON=1
This ensures experimental features are explicitly enabled by users
who want to try them.
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
Implements the channels infrastructure for connecting external messaging
platforms to Qwen Code via ACP. Phase 1 supports plain text round-trip:
Telegram user sends message -> AcpBridge -> qwen-code --acp -> response
back to Telegram.
New packages:
- @qwen-code/channel-base: AcpBridge, SessionRouter, SenderGate, ChannelBase
- @qwen-code/channel-telegram: TelegramAdapter using telegraf
CLI: `qwen channel start <name>` reads from settings.json channels config,
spawns ACP agent, connects to Telegram via polling.
* feat(storage): support configurable runtime output directory (#2014)
Add `advanced.runtimeOutputDir` setting and `QWEN_RUNTIME_DIR` env var
to redirect runtime output (temp files, debug logs, session data, todos,
insights) to a custom directory while keeping config files at ~/.qwen.
- Introduce `Storage.setRuntimeBaseDir()` / `getRuntimeBaseDir()` with
tilde expansion and relative path resolution
- Add `AsyncLocalStorage`-based `runWithRuntimeBaseDir()` for concurrent
session isolation in ACP integration
- Update all runtime path methods to use `getRuntimeBaseDir()` instead
of `getGlobalQwenDir()` (temp, debug, ide, projects, history dirs)
- Config paths (settings, oauth, installation_id, etc.) remain pinned
to `~/.qwen` regardless of runtime dir configuration
- Add comprehensive tests covering path resolution, env var priority,
async context isolation, and config path stability
* fix(core/storage): 支持 Windows 风格波浪号路径
扩展 setRuntimeBaseDir 以支持 Windows 风格的波浪号路径 (~\),
使用统一的路径分割逻辑处理 Unix 和 Windows 风格的路径分隔符
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
* fix(core/debugLogger): runtime base dir 变更时创建新 debug 目录
添加 ensuredDebugDirPath 追踪变量，当 runtime base dir 发生变更时，
确保在新的目录下创建 debug 子目录
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
* feat(cli/acp): 支持 ACP runtime output dir 配置
新增 runWithAcpRuntimeOutputDir 辅助函数，在 ACP Agent 的
loadSession 和 listSessions 操作中应用配置的 runtimeOutputDir
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
* docs(vscode-ide-companion/acpConnection): 补充 this 别名的使用说明
为 self = this 的用法添加解释性注释，说明在嵌套回调中需要使用 this
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
* feat(cli): add runtime output directory configuration support
* fix(core): update test to use getUserSkillsDirs method
Update storage.test.ts to call getUserSkillsDirs() instead of the
non-existent getUserSkillsDir() method. The method was renamed to
return an array of skill directories.
* fix(core/todoWrite): use path.join for cross-platform path assertion in test
Replace hardcoded forward-slash path `.qwen/todos/` with `path.join('.qwen', 'todos')` to fix Windows CI failure where paths use backslashes.
Made-with: Cursor
---------
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
- Add needsUtf8Bom() to detect when UTF-8 BOM is needed based on file
extension and system code page
- PowerShell 5.1 on non-UTF-8 Windows systems (e.g. GBK) requires BOM
to read scripts correctly
- Remove default UTF8 encoding; undefined now triggers auto-detection
- Add tests for needsUtf8Bom() covering Windows/non-Windows scenarios
This ensures PowerShell scripts are written with UTF-8 BOM on systems
that need it, fixing character encoding issues for non-ASCII content.
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
Remove the summarizeToolOutput setting and related functionality.
This feature allowed LLM-based summarization of shell tool output but is no longer needed.
This simplifies the codebase by removing unused summarization logic and configuration options.
Remove the enableToolOutputTruncation boolean setting. Users can now
disable truncation by setting truncateToolOutputThreshold to 0 or a
negative value instead of using a separate toggle.
This simplifies the configuration and prevents users from accidentally
disabling truncation which could cause memory issues with large tool
outputs.
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
The LLM-based loop detection was causing issues where the process would
stop unexpectedly. This change removes the LLM-based loop detection
feature entirely and sets skipLoopDetection to true by default to
prevent false positives from interrupting user sessions.
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
- Add arena_session_started, arena_agent_completed, arena_session_ended events
- Implement ArenaManager telemetry hooks with lifecycle tracking and metrics
- Update AgentStatistics to support API-provided totalTokenCount and remove estimatedCost
- Pass agent session IDs for telemetry correlation in PTY mode
This enables detailed observability into arena performance, agent
completion rates, and model comparison outcomes.
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
- SubAgentScope now automatically appends global memory (including QWEN.md and output-language.md) to its system prompt, ensuring subagents respect project conventions and language preferences.
- The CLI config loader now checks for `.qwen/output-language.md` in the project directory before falling back to the global `~/.qwen/output-language.md`, allowing project-specific language settings.
Made-with: Cursor
- Merge coder-model and qwen3.5-plus into a single coder-model with vision capability
- Remove vlmSwitchMode CLI argument and experimental.vlmSwitchMode setting
- Remove useVisionAutoSwitch hook and inline image format checking into useGeminiStream
- Remove ModelSwitchDialog and related vision switch UI components
- Update all related tests to reflect the simplified model structure
- Set DEFAULT_QWEN_MODEL to coder-model
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
Introduces a new Arena system for running multiple AI agents in parallel
terminal sessions with support for iTerm and Tmux backends.
Core:
- Add ArenaManager and ArenaAgentClient for orchestrating multi-agent sessions
- Add terminal backends (ITermBackend, TmuxBackend) with feature detection
- Add git worktree service for isolated agent workspaces
- Add arena event system for real-time status updates
CLI:
- Add /arena command with start, stop, status, and select subcommands
- Add Arena dialogs (Select, Start, Status, Stop)
- Add ArenaCards component for displaying parallel agent outputs
- Consolidate message components into StatusMessages and ConversationMessages
- Add MultiSelect component for agent selection
Config:
- Add arena-related settings to schema and config
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
