mirrors/hermes-agent
1
0
Fork 0
mirror of https://github.com/NousResearch/hermes-agent.git synced 2026-04-25 00:51:20 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix: recover partial streamed content on connection failure

Browse source 
When streaming fails after partial content delivery (e.g. OpenRouter
timeout kills connection mid-response), the stub response now carries
the accumulated streamed text instead of content=None.

Two fixes:
1. The partial-stream stub response includes recovered content from
   _current_streamed_assistant_text — the text that was already
   delivered to the user via stream callbacks before the connection
   died.

2. The empty response recovery chain now checks for partial stream
   content BEFORE falling back to _last_content_with_tools (prior
   turn content) or wasting API calls on retries. This prevents:
   - Showing wrong content from a prior turn
   - Burning 3+ unnecessary retry API calls
   - Falling through to '(empty)' when the user already saw content

The root cause: OpenRouter has a ~125s inactivity timeout. When
Anthropic's SSE stream goes silent during extended reasoning, the
proxy kills the connection. The model's text was already partially
streamed but the stub discarded it, triggering the empty recovery
chain which would show stale prior-turn content or waste retries.
This commit is contained in:
Teknium 2026-04-13 00:55:10 -07:00 • committed by Teknium
parent 35b11f48a5
commit 397eae5d93
2 changed files with 117 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

37
run_agent.py
View file

@ -5391,13 +5391,22 @@ class AIAgent:
# a new API call, creating a duplicate message. Return a
# partial "stop" response instead so the outer loop treats this
# turn as complete (no retry, no fallback).
# Recover whatever content was already streamed to the user.
# _current_streamed_assistant_text accumulates text fired
# through _fire_stream_delta, so it has exactly what the
# user saw before the connection died.
_partial_text = (
getattr(self, "_current_streamed_assistant_text", "") or ""
).strip() or None
logger.warning(
"Partial stream delivered before error; returning stub "
"response to prevent duplicate messages: %s",
"response with %s chars of recovered content to prevent "
"duplicate messages: %s",
len(_partial_text or ""),
result["error"],
)
_stub_msg = SimpleNamespace(
role="assistant", content=None, tool_calls=None,
role="assistant", content=_partial_text, tool_calls=None,
reasoning_content=None,
)
return SimpleNamespace(
@ -9889,6 +9898,30 @@ class AIAgent:
# Check if response only has think block with no actual content after it
if not self._has_content_after_think_block(final_response):
# ── Partial stream recovery ─────────────────────
# If content was already streamed to the user before
# the connection died, use it as the final response
# instead of falling through to prior-turn fallback
# or wasting API calls on retries.
_partial_streamed = (
getattr(self, "_current_streamed_assistant_text", "") or ""
)
if self._has_content_after_think_block(_partial_streamed):
_turn_exit_reason = "partial_stream_recovery"
_recovered = self._strip_think_blocks(_partial_streamed).strip()
logger.info(
"Partial stream content delivered (%d chars) "
"— using as final response",
len(_recovered),
)
self._emit_status(
"↻ Stream interrupted — using delivered content "
"as final response"
)
final_response = _recovered
self._response_was_previewed = True
break
# If the previous turn already delivered real content alongside
# tool calls (e.g. "You're welcome!" + memory save), the model
# has nothing more to say. Use the earlier content immediately