Problem

After upgrading to the latest version (2026.3.1), gateway restarts trigger deleteMyCommands + setMyCommands Telegram API calls on every boot. When the gateway is restarted a few times (e.g., during troubleshooting), this quickly hits Telegram's 429 rate limit with retry-after values of 460-490 seconds.

During the lockout period, the bot becomes completely unresponsive — messages show as "loading" in Telegram with no reply.

Expected Behavior

• Command sync should be throttled/debounced to avoid rate limits on repeated restarts
• If command sync fails with 429, the bot should still process incoming messages — command registration is non-critical
• Consider caching last-synced commands and skipping sync if unchanged

Observed Behavior

• Every openclaw gateway restart calls deleteMyCommands + setMyCommands
• After a few restarts, Telegram returns 429 Too Many Requests
• Bot is fully unresponsive during the lockout (~8 min)
• No messages are processed or replied to

Logs

telegram deleteMyCommands failed: Call to 'deleteMyCommands' failed! (429: Too Many Requests: retry after 491)
telegram setMyCommands failed: Call to 'setMyCommands' failed! (429: Too Many Requests: retry after 490)
Telegram command sync failed: GrammyError: Call to 'setMyCommands' failed! (429: Too Many Requests: retry after 490)

Environment

• OpenClaw: 2026.3.1
• Node: v22.22.0
• OS: macOS (Homebrew)
• Channel: Telegram
• Multiple agents on single gateway

Suggested Fix

1. Cache command hash, skip sync when unchanged
2. Decouple command sync failure from message polling
3. Add backoff for command sync retries