David/HERMES-GIELIE
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

fix: stop false Compressing-context card for preflight compression

Browse source 
PR #6184 narrowed the agent-status-to-SSE bridge to real compression
start notices, but the 'preflight compression:' pattern matches the
Hermes agent's preflight check message ("Preflight compression: ~X
tokens >= Y threshold"). This fires on every new session where the
system prompt alone exceeds the threshold, showing a "Compressing
context" card before any model output — a false positive.

Remove 'preflight compression:' from the pattern. The preflight check
is a one-time pre-model-call estimate, not a mid-conversation
compression event that the user needs to see.
This commit is contained in:
happy5318 2026-08-01 00:44:57 +08:00
parent dd7f6ac318
commit 9c1c1d8063
2 changed files with 5 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

3
api/streaming.py
View file

@ -587,8 +587,7 @@ def _is_agent_compression_start_status(kind: str, message: str) -> bool:
if 'compressed' in m and 'compressing' not in m and 'compression attempt' not in m: if 'compressed' in m and 'compressing' not in m and 'compression attempt' not in m:
return False return False
return ( return (
'preflight compression:' in m 'pre-api compression:' in m
or 'pre-api compression:' in m
or 'compacting context' in m or 'compacting context' in m
or 'context too large' in m or 'context too large' in m
or '— compressing (' in m or '— compressing (' in m

5
tests/test_auto_compression_card.py
View file

@ -473,7 +473,10 @@ def test_agent_status_callback_emits_compressing_and_warning_events():
def test_agent_compression_start_status_matches_real_emitters_only(): def test_agent_compression_start_status_matches_real_emitters_only():
# Real start notices from hermes-agent emitters # Real start notices from hermes-agent emitters
assert _is_agent_compression_start_status( # Preflight compression is excluded — it fires on every new session
# where the system prompt alone exceeds the threshold, producing a
# false "Compressing context" card before any model output.
assert not _is_agent_compression_start_status(
"lifecycle", "lifecycle",
"📦 Preflight compression: ~101,000 tokens >= 96,000 threshold. This may take a moment.", "📦 Preflight compression: ~101,000 tokens >= 96,000 threshold. This may take a moment.",
) )