fix(core): always trigger done_emoji when agent finishes

Previously done_emoji only fired when stream preview had been updated
in-place at least once, which meant quick responses never triggered
the reaction. Remove the needsDoneReaction() gate so the platform's
AddDoneReaction is always called on EventResult.

Made-with: Cursor

config.example.toml

@@ -746,8 +746,8 @@ app_secret = "your-feishu-app-secret"
 #                              # 允许的用户 open_id，逗号分隔；"*" 表示所有（默认）。发送 /whoami 获取你的 open_id。
 # reaction_emoji = "OnIt"  # Emoji reaction on incoming messages (default: "OnIt"); set to "none" to disable
 #                           # 收到消息时添加的表情回复（默认 "OnIt"）；设为 "none" 可禁用
-# done_emoji = "none"       # Emoji reaction after agent finishes streaming work in quiet mode (e.g. "Done"); set to "none" to disable
-#                           # Agent 在 quiet 模式下完成流式工作后的表情回复（如 "Done"）；设为 "none" 可禁用
+# done_emoji = "none"       # Emoji reaction when agent finishes a reply (e.g. "Done"); set to "none" to disable
+#                           # Agent 完成回复后添加的表情回复（如 "Done"）；设为 "none" 可禁用
 # group_reply_all = false   # If true, respond to ALL group messages without requiring @mention
 #                           # 设为 true 时，群聊中无需 @机器人 也会响应所有消息（默认 false）
 # share_session_in_channel = false  # If true, all users in a group share one agent session

core/engine.go

@@ -3080,15 +3080,11 @@ func (e *Engine) processInteractiveEvents(state *interactiveState, session *Sess
 				}
 			}
 
-			// When the preview was updated in-place at least once (e.g. Feishu card
-			// streaming), the user only saw a push for the initial send and subsequent
-			// updates were silent. Add a "done" reaction so they know the agent finished.
-			// The reaction is added after stopTyping (deferred) so the doing emoji
-			// is removed first.
-			if sp.needsDoneReaction() {
-				if doneTI, ok := p.(TypingIndicatorDone); ok {
-					doneReaction = func() { doneTI.AddDoneReaction(replyCtx) }
-				}
+			// Add a "done" reaction so the user knows the agent finished.
+			// The reaction is added after stopTyping (deferred) so the
+			// "doing" emoji is removed first.
+			if doneTI, ok := p.(TypingIndicatorDone); ok {
+				doneReaction = func() { doneTI.AddDoneReaction(replyCtx) }
 			}
 
 			return

docs/feishu.md

@@ -111,14 +111,14 @@ app_secret = "QhkMpxxxxxxxxxxxxxxxxxxxx"
 # enable_feishu_card = true  # 可选：关闭后统一回退纯文本回复
 # thread_isolation = true    # 可选：按飞书 thread/root 隔离群聊会话
 # progress_style = "legacy"  # 可选：legacy | compact | card
-# done_emoji = "none"          # 可选：quiet 模式下 agent 流式工作完成后添加的表情回复（如 "Done"）；设为 "none" 可禁用
+# done_emoji = "none"          # 可选：agent 完成回复后添加的表情回复（如 "Done"）；设为 "none" 可禁用
 ```
 
 > 如果应用没有交互卡片权限，或后台未配置卡片回调，可将 `enable_feishu_card = false`，让所有命令统一走纯文本回复，避免卡片发送失败后用户看不到内容。
 > 如果开启 `thread_isolation = true`，群聊里每个根消息 / reply thread 会对应一个独立 agent session；私聊行为保持原样。
 > `progress_style = "compact"` 会把思考/工具进度合并到一条可更新消息里，减少刷屏；`legacy` 保持原有逐条发送；`card` 会使用结构化卡片（标题 + 进度块）持续更新同一条消息，观感比纯文本更清晰。
 > `domain` 只影响运行时 API / WebSocket 请求地址；CLI `setup/new/bind` 的引导域名仍然使用内置默认值。
-> `done_emoji` 仅在 quiet 模式下、飞书卡片被流式更新过（通过 UpdateMessage 原地编辑）时生效：先移除 "OnIt" 表情，再添加你指定的 done 表情，这样用户能收到飞书推送通知。飞书卡片原地更新本身不会触发推送，所以 quiet 模式下用户可能不知道 agent 已完成。单轮纯文本回复（只通过 SendPreviewStart 发送）不会触发此行为。设为 `"none"` 或不配置则禁用。
+> `done_emoji` 设置后，agent 每次完成回复时会在用户消息上添加指定表情（如 `"Done"` → ✅）。先移除 "OnIt" 表情（如果有），再添加 done 表情。在 quiet 模式下特别有用，因为飞书卡片原地更新不触发推送，done 表情可以通知用户 agent 已完成。设为 `"none"` 或不配置则禁用。
 
 ---