fix: improve platform gateway reliability — multi-process protection, hook chat_id, and Feishu event queueing

[milkoor]

Problem

Platform gateways (Feishu, Telegram, etc.) suffer from three reliability issues:

1. Silent message drops: When the Feishu WebSocket callback fires before the main event loop is ready, inbound messages are dropped with a warning log but never delivered.
2. Concurrent gateway processes: Multiple gateway processes (e.g. hermes run + hermes gateway) share the lark_oapi global state, meaning only the last-started process receives events. This causes silent disconnections where Feishu shows connected but messages go unanswered.
3. Missing chat_id in hook context: Gateway hooks (e.g. card-streaming hooks) receive session_id but not chat_id, preventing platform adapters from creating platform-specific cards/messages without using internal session IDs.

Changes

gateway/run.py

• Add chat_id to hook_ctx for agent:start events, enabling hook handlers to target the correct platform destination

hermes_cli/gateway.py

• Prevent concurrent gateway instances by checking for existing gateway processes before starting
• When an existing instance is detected, the new instance aborts with a helpful message (hermes gateway run --replace to override)

gateway/platforms/feishu.py

• Queue inbound Feishu WebSocket events when the adapter loop is not yet ready, then replay them once the loop is available
• Previously these events were silently dropped, causing users to lose their first message after gateway restart

Testing

• ✅ All three patches apply cleanly against latest main (cc54818)
• ✅ Gateway syntax check passes for all modified files
• ✅ Tested on live Feishu gateway — messages flow correctly after restart

[britrik]

Code Review: PR #4789

Summary

The PR introduces three reliability improvements:

1. Feishu event queueing - queues events when the adapter loop isn't ready
2. chat_id hook addition - adds chat_id to the agent:start hook context
3. Multi-process protection - prevents duplicate gateway instances

---

Issues & Suggestions

1. Feishu Event Queueing (gateway/platforms/feishu.py)

Issue: Thread spawning per event is wasteful

Every inbound message event before the loop is ready spawns a new thread. If 10 events arrive quickly, that's 10 threads. Consider using a single background worker or batching.

Issue: No bounds checking on _pending_events
If the loop never becomes ready (e.g., prolonged startup issue), the list grows unbounded, potentially causing memory exhaustion. Consider adding a max queue size.

Issue: Silent failure in _drain_pending_events
The call in has no done_callback to log failures. If an event fails to process, it silently disappears.

Suggestion: Add a try/except around the coroutine call:

---

2. Multi-Process Protection (hermes_cli/gateway.py)

Issue: Fragile environment variable check

This could be bypassed if another process sets this env var. Consider using a lock file or a more robust mechanism like PID files with validation.

Positive: The find_gateway_pids() function is solid and handles both Linux/macOS and Windows.

---

3. chat_id Hook Addition (gateway/run.py)

Looks good: Straightforward addition. The is used extensively elsewhere in the codebase (lines 1717, 1727, 2477, etc.), so this should work correctly.

---

Overall

The changes address real reliability issues. The Feishu queueing could be improved with a single worker thread instead of per-event threads, and failure handling could be better. The multi-process protection and chat_id hook are solid.