mirrors/hermes-agent
1
0
Fork 0
mirror of https://github.com/NousResearch/hermes-agent.git synced 2026-04-29 01:31:41 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 6

fix: audit fixes — 5 bugs found and resolved

Browse source 
Thorough code review found 5 issues across run_agent.py, cli.py, and gateway/:

1. CRITICAL — Gateway stream consumer task never started: stream_consumer_holder
   was checked BEFORE run_sync populated it. Fixed with async polling pattern
   (same as track_agent).

2. MEDIUM-HIGH — Streaming fallback after partial delivery caused double-response:
   if streaming failed after some tokens were delivered, the fallback would
   re-deliver the full response. Now tracks deltas_were_sent and only falls
   back when no tokens reached consumers yet.

3. MEDIUM — Codex mode lost on_first_delta spinner callback: _run_codex_stream
   now accepts on_first_delta parameter, fires it on first text delta. Passed
   through from _interruptible_streaming_api_call via _codex_on_first_delta
   instance attribute.

4. MEDIUM — CLI close-tag after-text bypassed tag filtering: text after a
   reasoning close tag was sent directly to _emit_stream_text, skipping
   open-tag detection. Now routes through _stream_delta for full filtering.

5. LOW — Removed 140 lines of dead code: old _streaming_api_call method
   (superseded by _interruptible_streaming_api_call). Updated 13 tests in
   test_run_agent.py and test_openai_client_lifecycle.py to use the new
   method name and signature.

4573 tests passing.
This commit is contained in:
teknium1 2026-03-16 06:35:46 -07:00
parent 99369b926c
commit 8e07f9ca56
5 changed files with 75 additions and 176 deletions
Download patch file Download diff file Expand all files Collapse all files

11
tests/test_openai_client_lifecycle.py
View file

@ -59,8 +59,11 @@ def _build_agent(shared_client=None):
agent._interrupt_requested = False
agent._interrupt_message = None
agent._client_lock = threading.RLock()
agent._client_kwargs = {"api_key": "test-key", "base_url": agent.base_url}
agent._client_kwargs = {"api_key": "***", "base_url": agent.base_url}
agent.client = shared_client or FakeSharedClient(lambda **kwargs: {"shared": True})
agent.stream_delta_callback = None
agent._stream_callback = None
agent.reasoning_callback = None
return agent
@ -173,7 +176,11 @@ def test_streaming_call_recreates_closed_shared_client_before_request(monkeypatc
monkeypatch.setattr(run_agent, "OpenAI", factory)
agent = _build_agent(shared_client=stale_shared)
response = agent._streaming_api_call({"model": agent.model, "messages": []}, lambda _delta: None)
agent.stream_delta_callback = lambda _delta: None
# Force chat_completions mode so the streaming path uses
# chat.completions.create(stream=True) instead of Codex responses.stream()
agent.api_mode = "chat_completions"
response = agent._interruptible_streaming_api_call({"model": agent.model, "messages": []})
assert response.choices[0].message.content == "Hello world"
assert agent.client is replacement_shared