openclaw status --deep Channels table omits Feishu when Feishu is configured and connected

Summary

On OpenClaw 2026.5.3-1 (macOS arm64, Node 25.8.1), the Channels table rendered by openclaw status --deep only lists Telegram, even though channels.feishu is fully configured (10 enabled accounts via WebSocket connection mode), the Feishu plugin successfully started in the previous session, and inbound Feishu messages route to agents normally.

This is a display-only bug — Feishu itself is healthy — but it caused real incident-response confusion (the operator initially thought Feishu was disconnected because the deep-status table appeared to omit it).

Environment

• OpenClaw: 2026.5.3-1 (pnpm install, npm latest)
• OS: macOS 26.4.1 (arm64)
• Node: v25.8.1
• Channels configured: channels.telegram, channels.feishu
• Feishu: 10 agent accounts + 1 default account, all enabled: true, connectionMode: websocket, appId/appSecret set
• Gateway: local loopback, LaunchAgent installed, running

Repro

1. Configure channels.feishu with at least one enabled account (websocket mode).
2. Start gateway, confirm Feishu WebSocket connections succeed in ~/.openclaw/logs/gateway.log (e.g. [feishu] feishu[<account>]: WebSocket client started and bot open_id recovered via background retry).
3. Send a Feishu DM to any configured agent — it routes correctly and the agent responds.
4. Run openclaw status --deep.

Expected

The Channels table lists both Telegram and Feishu, with Feishu showing per-account state (e.g. accounts 11/11).

Actual

The Channels table lists only Telegram:

Channels
┌──────────┬─────────┬────────┬─────────────────────────────────────────────────┐
│ Channel  │ Enabled │ State  │ Detail                                          │
├──────────┼─────────┼────────┼─────────────────────────────────────────────────┤
│ Telegram │ ON      │ OK     │ token config (… · len 46) · accounts 1/1        │
└──────────┴─────────┴────────┴─────────────────────────────────────────────────┘

The Security audit section in the same status --deep run does reference channels.feishu.accounts.default.groupPolicy, confirming the runtime sees Feishu config. The Health section only lists Telegram. So Feishu is loaded by the runtime but never rendered into the Channels/Health summary tables.

Evidence Feishu is actually live

• Feishu WebSocket logs show all 10 lobster accounts plus default started cleanly:

  [feishu] feishu[market_lobster]: WebSocket client started
  [feishu] feishu[decision_lobster]: bot open_id recovered via background retry: ou_…

  (no disconnect / reconnect / error since the boot at 2026-05-03 09:03 PDT)
• Inbound Feishu messages reach the right agent and produce normal responses (sessions appear as agent:<id>:feishu:direct:<openId> in the Sessions table).
• gateway config.get channels.feishu.enabled returns true; full channels.feishu block is intact in the merged runtime config.

Likely cause

The Channels/Health renderer in status --deep appears to only enumerate channels that surfaced through the legacy plugin loader, not channels declared as built-in under channels.*. After Feishu was promoted from plugins.entries.feishu to a built-in channels.feishu block, the deep-status renderer was not updated to enumerate it.

A related symptom: every config reload still emits

- plugins.entries.feishu: plugin not found: feishu (stale config entry ignored; remove it from plugins config)

even though plugins.entries.feishu is not present in ~/.openclaw/openclaw.json. Something inside the runtime keeps re-injecting/expecting the legacy plugins.entries.feishu shape, and gateway config.patch to clear it is rejected with cannot change protected config paths: plugins.entries.feishu.enabled. This may be the same code path that the deep-status renderer relies on to discover Feishu.

Impact

• Operator confusion during incidents — looks like Feishu died after upgrade.
• Misleading health/probe surface for Feishu-primary deployments.

Suggested fix

1. Update the status --deep Channels and Health renderers to enumerate built-in channels under channels.* (Feishu, and any other promoted-from-plugin channels), not only legacy plugin-discovered channels.
2. Either persist the stale plugins.entries.feishu warning suppression once the plugin entry is gone, or unblock gateway config.patch for plugins.entries.feishu.* so operators can clean it up themselves.

Workaround

None needed for actual messaging — Feishu works. For visibility, operators can verify Feishu via:

• gateway config.get channels.feishu.enabled
• grep '\[feishu\]' ~/.openclaw/logs/gateway.log | tail -20
• inbound message routing to agent:<id>:feishu:direct:<openId> sessions in openclaw status Sessions table.

Happy to provide the full sanitized status --deep output, gateway log excerpts, or the merged runtime config for channels.feishu if it helps.