Generates natural private-chat greetings and idle follow-ups with a selected LLM.
Proactive Companion for LangBot
Generates scheduled greetings and idle follow-ups for one private-chat recipient with a LangBot-configured LLM. It preserves the active persona, recent dialogue, and relevant memory context while applying quiet hours, cooldowns, daily limits, and an unanswered-message guard.
Requirements
- LangBot 4.3 or later with a working private-message platform adapter (the current release targets QQ/NapCat).
- One chat model configured in LangBot.
- The target user must message the bot once after installation so the plugin can capture a valid delivery context.
- Optional: LongTermMemory before this plugin in the pipeline; Irodori TTS for proactive voice delivery.
Installation and setup
- Install from LangBot Marketplace or a
.lbpkgrelease. - Add Proactive Companion to the target pipeline. Place it after LongTermMemory when used.
- Select the sending bot, recipient QQ, generation model, and time zone.
- Keep the master switch off while reviewing schedule, limits, and privacy.
- Let the recipient send one private message, then enable the plugin.
Morning, noon, evening, and idle paths share the daily maximum, cooldown, quiet period, and unanswered limit. Each scheduled window chooses one candidate time per day; it does not reroll every minute.
Context and storage
The selected model may receive the current persona, a bounded recent-dialogue snapshot, local time, and relevant memory exposed by the pipeline. Only counters, timestamps, scheduling state, and delivery context are kept through LangBot plugin storage; no second full transcript is persisted.
!reset clears LangBot's normal conversation session. The plugin also observes reset events and clears its recent-context cache so later proactive messages do not revive the previous topic.
Safety defaults
- Master switch is off by default.
- Only one configured private-chat recipient is supported.
- New user input cancels an in-flight proactive send.
- By default, no second proactive message is sent until the user replies.
- Model failure produces no fabricated fallback message.
See PRIVACY.md, THIRD_PARTY_NOTICES.md, and the Simplified Chinese guide.
Uninstall
Disable the plugin, remove it from every pipeline, then uninstall it in LangBot.
License
MIT. See LICENSE.
Use Proactive Companion in Feishu/Lark bots, DingTalk bots, WeCom bots, WeChat bots, QQ bots, Slack bots, Discord bots, Telegram bots, LINE Bots, and Matrix Bots.
Host your bot on LangBot Cloud and connect this extension to the collaboration platforms your team already uses.