gooneyryan/hermes-agent
0
0
Fork 0
mirror of https://github.com/NousResearch/hermes-agent.git synced 2026-05-21 03:39:54 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(copilot): GitHub Models 413 hint — port to extracted conversation_loop

Browse Source 
Original commits 4ded3ede3 (@konsisumer) + 374dc81c2 (Teknium) added a
413 hint to run_agent.py's agent loop. Final-state version (the sharpened
374dc81c2 wording) ported to agent/conversation_loop.py, where the
payload_too_large branch now lives.

The deprecation detection + _URL_TO_PROVIDER changes from both commits
landed in agent/copilot_acp_client.py and agent/model_metadata.py via
the prior merge.

Closes #10648

Co-authored-by: konsisumer <der@konsi.org>
Co-authored-by: Teknium <127238744+teknium1@users.noreply.github.com>
This commit is contained in:
teknium1
2026-05-16 23:38:45 -07:00
parent 3fbedd732e
commit df22d29522
1 changed files with 33 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
agent/conversation_loop.py
+33
View File
@@ -2333,6 +2333,39 @@ def run_conversation(
classified.reason == FailoverReason.payload_too_large
)
# Actionable hint for GitHub Models (Azure) 413 errors.
# The free tier enforces a hard 8K token cap per request,
# which Hermes' system prompt + tool schemas alone exceed.
# Compression can't help — the floor is the system prompt
# itself, not the conversation — so surface a clear "not
# compatible" message instead of looping into three futile
# compression attempts.
if (
status_code == 413
and isinstance(agent.base_url, str)
and "models.inference.ai.azure.com" in agent.base_url
):
agent._vprint(
f"{agent.log_prefix} 💡 GitHub Models free tier (models.inference.ai.azure.com) caps every",
force=True,
)
agent._vprint(
f"{agent.log_prefix} request at ~8K tokens. Hermes' system prompt + tool schemas baseline",
force=True,
)
agent._vprint(
f"{agent.log_prefix} exceeds that floor, so this endpoint cannot run an agentic loop.",
force=True,
)
agent._vprint(
f"{agent.log_prefix} Use the `copilot` provider with a Copilot subscription token (`hermes",
force=True,
)
agent._vprint(
f"{agent.log_prefix} setup` → GitHub Copilot), or pick any other provider.",
force=True,
)
if is_payload_too_large:
compression_attempts += 1
if compression_attempts > max_compression_attempts: