A deleted or archived Slack channel leaves its room bound, agents posting into nothing, and no one told

オープン
#2,001 コメント 1 件 リアクション 0 件 担当者 0 名 GitHub で見る

メンテナーはふだん 1 日以内に返信

@Darshan3690 がすでに取り組んでいます。

2026年11月15日 から。

評価

難易度
3/5
見積もり時間
半日
初心者へのやさしさ
45/100
issue の種類
バグ
明瞭さ
おおむね明確
活発さ
活発
技術スタック
python
領域
backend

調査の方向性

Start in core/switch_core/bridges/collaboration/slack/adapter.py: dispatch_event falls through with no branch for channel_deleted, channel_archive or channel_unarchive, and send_message swallows SlackApiError after logger.error. The manifest that requests the event subscriptions is also relevant. Done means channel_not_found and is_archived are handled as terminal, with a visible system line posted to the room; the manifest change is optional and larger.

索引モデルが issue の本文から書いたものです。

説明

What happens today

Delete (or archive) a Slack channel that a Switch room is bridged to, and Switch does not react at all.

  • The event never arrives. The app manifest requests app_home_opened, message.channels, message.groups, message.im and message.mpim. channel_deleted, channel_archive and channel_unarchive are not among them.
  • It would not be handled if it did. dispatch_event in core/switch_core/bridges/collaboration/slack/adapter.py routes slash commands, interactive payloads, message and member_joined_channel. Anything else falls off the end of the function with no branch and no log.
  • The room stays bound. bridge_id, external_channel_id and channel_type keep pointing at a channel that no longer exists, and the in-memory map is rebuilt from those rows on every restart. The only code that clears a binding is whole-workspace disconnect.
  • Agents carry on. An agent's post_message succeeds and persists, because the relay to Slack happens afterwards in a separate consumer. send_message then swallows the SlackApiError — logger.error(...) and return None — so channel_not_found and is_archived are indistinguishable from a transient failure and from each other.
  • Nobody is told. No system post in the room, no notice to a workspace admin, no telemetry event. The only durable signal is repeated error lines in the server log.

Archiving behaves identically. If someone later un-archives the channel the bridge silently resumes, because the stale mapping is still intact — accidental rather than designed.

Why this is worth fixing

This is the failure mode CLAUDE.md singles out: "Prefer a visible failure over a silent fallback" and "Never silently swallow errors to keep things 'working.'" A room whose external half has been deleted looks completely healthy from inside Switch. People go on addressing agents, agents go on answering, and none of it reaches anyone.

There is a second-order consequence worth noting separately: messages already mirrored from that channel into Switch are retained indefinitely, since message rows cascade on the room foreign key and the room is never archived or deleted. Deleting a channel in Slack therefore removes the Slack-side copy and nothing else. That may well be the correct product behaviour, but it is currently undocumented and nothing surfaces it.

Suggested direction

Smallest useful change first:

  1. Distinguish the dead-channel errors from transient ones. channel_not_found and is_archived are terminal; treating them as a generic send failure is what makes this invisible.
  2. Say so in the room. A system line ("this room's Slack channel no longer exists; messages are not reaching it") turns a silent failure into a visible one and needs no new subscription.
  3. Optionally subscribe to channel_deleted / channel_archive to react immediately rather than on the next failed send. This needs a manifest change, so it is the larger step.

Marking the room as unbridged automatically is deliberately not suggested here — that destroys the mapping an un-archive would otherwise restore, and it is a product decision rather than a bug fix.

主要言語
Python
スター
4.5k
フォーク
348
平均マージ
1日 17時間
マージ済み PR(30日)
287

環境構築

はじめの一歩

  1. issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
  2. 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
  3. リポジトリをフォークし、ブランチを切って変更します。
  4. issue 番号を参照したプルリクエストを送ります。

sandbox-quantum/switch のほかの issue

sandbox-quantum/switch の issue をすべて見る

似ている issue

Python の issue をもっと見る

新しい issue をメールで受け取る

初心者向けの GitHub issue を短くまとめたダイジェスト。