fix(daemon): enable auto-title generation for ACP sessions#4836
Conversation
The automatic session title generation was silently disabled for all daemon sessions. The guard in `maybeTriggerAutoTitle` checks `config.isInteractive()`, which returns false for the ACP child process because it is spawned with pipe stdio (`process.stdin.isTTY === false`). This guard was originally added to prevent headless one-shot CLI runs (`qwen -p "do something"`) from wasting fast-model tokens on a title that no one would ever see. However, daemon sessions are long-lived and user-resumable — they appear in the session list and benefit from semantically meaningful titles. The fix allows ACP mode (`config.getExperimentalZedIntegration()`) to bypass the `isInteractive()` check while still blocking true headless CLI runs. After this change, the first assistant reply in a daemon session will trigger LLM-based title generation (3-7 words, sentence case) using the configured fast model, just as it does for interactive TUI sessions. Generated with AI Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
|
Thanks for the PR @chiga0! Template: The PR uses custom headings (Background, Summary, etc.) instead of the template sections. Notably missing: Reviewer Test Plan (How to verify + Evidence), Tested on table, Risk & Scope, Linked Issues, and the Chinese translation. The content in Background/Summary is solid, but a reviewer needs the structured test plan to verify efficiently. Could you update the body to match the template? On direction: solid bug fix. Daemon sessions are long-lived and resumable, so skipping auto-title is a real gap. The On approach: minimal and well-scoped. The fix is a single condition relaxation — Template is the only blocker — once that's fixed, moving on to code review. 🔍 中文说明感谢贡献 @chiga0！ 模板： PR 使用了自定义标题（Background、Summary 等），而非模板要求的章节。缺少：Reviewer Test Plan（How to verify + Evidence）、Tested on 表格、Risk & Scope、Linked Issues，以及中文翻译。Background/Summary 的内容本身不错，但审查者需要结构化的测试计划来高效验证。能否按模板更新 PR 描述？ 方向：合理的 bug 修复。Daemon 会话是长期存在且可恢复的，跳过自动标题确实是一个真实的缺口。 方案：最小化且范围合理。修复是一个单一的条件放宽—— 模板是唯一的阻塞项——修复后进入代码审查。🔍 — Qwen Code · qwen3.7-max |
qwen-code-ci-bot
left a comment
There was a problem hiding this comment.
Thanks for the PR @chiga0!
Template: The PR uses custom headings (Background, Summary, etc.) instead of the template sections. Notably missing: Reviewer Test Plan (How to verify + Evidence), Tested on table, Risk & Scope, Linked Issues, and the Chinese translation. The content in Background/Summary is solid, but a reviewer needs the structured test plan to verify efficiently. Could you update the body to match the template?
On direction: solid bug fix. Daemon sessions are long-lived and resumable, so skipping auto-title is a real gap. The isInteractive() guard was designed for one-shot CLI runs, and ACP sessions legitimately bypass that assumption. Claude Code's CHANGELOG has related session-title fixes (e.g. "Fixed session title being generated from plugin monitor output"), confirming this is an active concern across agent CLIs. ✅
On approach: minimal and well-scoped. The fix is a single condition relaxation — !isInteractive() && !getExperimentalZedIntegration() — which preserves the original intent (don't waste tokens on throwaway sessions) while carving out the daemon case. The truth table in the PR body makes the logic easy to verify. 2 tests covering the new branch + the preserved guard. Nothing to cut. ✅
Template is the only blocker — once that's fixed, moving on to code review. 🔍
中文说明
感谢贡献 @chiga0！
模板： PR 使用了自定义标题（Background、Summary 等），而非模板要求的章节。缺少：Reviewer Test Plan（How to verify + Evidence）、Tested on 表格、Risk & Scope、Linked Issues，以及中文翻译。Background/Summary 的内容本身不错，但审查者需要结构化的测试计划来高效验证。能否按模板更新 PR 描述？
方向：合理的 bug 修复。Daemon 会话是长期存在且可恢复的，跳过自动标题确实是一个真实的缺口。isInteractive() 守卫是为一次性 CLI 运行设计的，ACP 会话合理地绕过了这个假设。Claude Code 的 CHANGELOG 也有相关的会话标题修复，确认这是 agent CLI 的共同关注点。✅
方案：最小化且范围合理。修复是一个单一的条件放宽——!isInteractive() && !getExperimentalZedIntegration()——保留了原始意图（不在一次性会话上浪费 token），同时为 daemon 场景开了口子。PR 描述中的真值表让逻辑一目了然。2 个测试覆盖新分支和保留的守卫。没有多余的部分。✅
模板是唯一的阻塞项——修复后进入代码审查。🔍
— Qwen Code · qwen3.7-max
qwen-code-ci-bot
left a comment
There was a problem hiding this comment.
No issues found. LGTM! ✅ Downgraded from Approve to Comment: CI still running. — qwen3.7-max via Qwen Code /review
|
@qwen-code /review |
qwen-code-ci-bot
left a comment
There was a problem hiding this comment.
No issues found. LGTM! ✅ — qwen3.7-max via Qwen Code /review
wenshao
left a comment
There was a problem hiding this comment.
No issues found. LGTM! ✅ — qwen3.7-max via Qwen Code /review
What this PR does
Allow ACP (daemon) sessions to auto-generate session titles after the first assistant reply, matching the existing behavior of interactive TUI sessions. The fix relaxes a single guard in
maybeTriggerAutoTitle()fromif (!isInteractive()) returntoif (!isInteractive() && !getExperimentalZedIntegration()) return, so daemon sessions are no longer treated as headless one-shot CLI runs.Why it's needed
The
isInteractive()guard was originally added to prevent headless CLI runs (qwen -p "...") from wasting fast-model tokens on a session title nobody would see. However, the ACP child process is spawned with pipe stdio (process.stdin.isTTY === false), which causesisInteractive()to returnfalsefor ALL daemon sessions — not just headless ones. As a result, every daemon session (Web Chat, IDE extension, Channel) shows truncated raw prompt text in the session list instead of a semantically meaningful title.isInteractive()getExperimentalZedIntegration()truefalsefalsetruefalsefalseReviewer Test Plan
How to verify
chatRecordingService.ts:955-965, verify the condition!isInteractive() && !getExperimentalZedIntegration()correctly skips only headless CLI mode.npx vitest run packages/core/src/services/chatRecordingService.autoTitle.test.ts— all 17 tests pass (15 existing + 2 new).isInteractive() === false; (b) headless CLI mode (non-interactive + non-ACP) still blocks.Evidence (Before & After)
Non-UI change — unit tests only.
Before:
maybeTriggerAutoTitlereturns early for any non-interactive config, including ACP.After:
maybeTriggerAutoTitleproceeds for ACP sessions (getExperimentalZedIntegration() === true).Tested on
Risk & Scope
AUTO_TITLE_ATTEMPT_CAP = 3.Linked Issues
None — discovered during daemon session lifecycle analysis.
中文说明
本 PR 做了什么
允许 ACP（daemon）会话在第一个 assistant 回复后自动生成会话标题，与交互式 TUI 会话的现有行为一致。修复方式是将
maybeTriggerAutoTitle()中的守卫条件从if (!isInteractive()) return放宽为if (!isInteractive() && !getExperimentalZedIntegration()) return，使 daemon 会话不再被当作一次性无头 CLI 运行。为什么需要
isInteractive()守卫最初是为了防止无头 CLI 运行（qwen -p "..."）在不会被看到的会话标题上浪费 fast model token。但 ACP 子进程通过 pipe stdio 启动（process.stdin.isTTY === false），导致所有 daemon 会话的isInteractive()返回false——不仅仅是无头运行。结果是所有 daemon 会话（Web Chat、IDE 扩展、Channel）在会话列表中显示截断的原始 prompt 文本，而非语义化的标题。风险与范围
AUTO_TITLE_ATTEMPT_CAP = 3限制。🤖 Generated with Qwen Code