Timeouts and Error Handling Improvements | Generated by AI

Home 2026.05 English 中文

---

Done. Here’s a summary of all changes across 3 files:

1. /Users/lzwjava/projects/ww/ww/llm/openrouter_client.py

• Added timeout=(5, 30) to requests.post() — 5s connect, 30s read (was: no timeout = hang forever)
• Same fix for the streaming endpoint

2. /Users/lzwjava/projects/ww/ww/note/create_note_from_clipboard.py

• Replaced sys.exit(1) with raise ValueError(...) for empty/short content (was: killed the entire Hermes CLI)
• Removed unused sys import

3. /Users/lzwjava/projects/ww/ww/note/create_note_utils.py

• Replaced sys.exit(1) with raise RuntimeError(...) in _call_llm_or_exit() (was: killed the CLI on LLM failure)
• Removed unused traceback and sys imports

4. /Users/lzwjava/.hermes/plugins/note/__init__.py

• Added import os
• Added fallback MODEL env var loading from ~/projects/ww/.env when dotenv doesn’t populate it
• Added print("Generating title via LLM...") so user sees feedback
• Added ValueError catch for content validation errors (returns clean message instead of traceback)

Try running /note again — you should now see “Generating title via LLM…” immediately, and if the API is slow it’ll timeout after 30s instead of hanging forever.

---

Back Donate