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: resolve three high-impact community bugs (#5819, #6893, #3388) (#7881)

Browse source 
Matrix gateway: fix sync loop never dispatching events (#5819)
- _sync_loop() called client.sync() but never called handle_sync()
  to dispatch events to registered callbacks — _on_room_message was
  registered but never fired for new messages
- Store next_batch token from initial sync and pass as since= to
  subsequent incremental syncs (was doing full initial sync every time)
- 17 comments, confirmed by multiple users on matrix.org

Feishu docs: add interactive card configuration for approvals (#6893)
- Error 200340 is a Feishu Developer Console configuration issue,
  not a code bug — users need to enable Interactive Card capability
  and configure Card Request URL
- Added required 3-step setup instructions to feishu.md
- Added troubleshooting entry for error 200340
- 17 comments from Feishu users

Copilot provider drift: detect GPT-5.x Responses API requirement (#3388)
- GPT-5.x models are rejected on /v1/chat/completions by both OpenAI
  and OpenRouter (unsupported_api_for_model error)
- Added _model_requires_responses_api() to detect models needing
  Responses API regardless of provider
- Applied in __init__ (covers OpenRouter primary users) and in
  _try_activate_fallback() (covers Copilot->OpenRouter drift)
- Fixed stale comment claiming gateway creates fresh agents per message
  (it caches them via _agent_cache since the caching was added)
- 7 comments, reported on Copilot+Telegram gateway
This commit is contained in:
Teknium 2026-04-11 11:12:20 -07:00 • committed by GitHub
parent 69f3aaa1d6
commit 06e1d9cdd4
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
5 changed files with 95 additions and 14 deletions
Download patch file Download diff file Expand all files Collapse all files

34
gateway/platforms/matrix.py
View file

@ -427,6 +427,11 @@ class MatrixAdapter(BasePlatformAdapter):
if isinstance(sync_data, dict):
rooms_join = sync_data.get("rooms", {}).get("join", {})
self._joined_rooms = set(rooms_join.keys())
# Store the next_batch token so incremental syncs start
# from where the initial sync left off.
nb = sync_data.get("next_batch")
if nb:
await client.sync_store.put_next_batch(nb)
logger.info(
"Matrix: initial sync complete, joined %d rooms",
len(self._joined_rooms),
@ -818,19 +823,40 @@ class MatrixAdapter(BasePlatformAdapter):
async def _sync_loop(self) -> None:
"""Continuously sync with the homeserver."""
client = self._client
# Resume from the token stored during the initial sync.
next_batch = await client.sync_store.get_next_batch()
while not self._closing:
try:
sync_data = await self._client.sync(timeout=30000)
sync_data = await client.sync(
since=next_batch, timeout=30000,
)
if isinstance(sync_data, dict):
# Update joined rooms from sync response.
rooms_join = sync_data.get("rooms", {}).get("join", {})
if rooms_join:
self._joined_rooms.update(rooms_join.keys())
# Share keys periodically if E2EE is enabled.
if self._encryption and getattr(self._client, "crypto", None):
# Advance the sync token so the next request is
# incremental instead of a full initial sync.
nb = sync_data.get("next_batch")
if nb:
next_batch = nb
await client.sync_store.put_next_batch(nb)
# Dispatch events to registered handlers so that
# _on_room_message / _on_reaction / _on_invite fire.
try:
await self._client.crypto.share_keys()
tasks = client.handle_sync(sync_data)
if tasks:
await asyncio.gather(*tasks)
except Exception as exc:
logger.warning("Matrix: sync event dispatch error: %s", exc)
# Share keys periodically if E2EE is enabled.
if self._encryption and getattr(client, "crypto", None):
try:
await client.crypto.share_keys()
except Exception as exc:
logger.warning("Matrix: E2EE key share failed: %s", exc)

37
run_agent.py
View file

@ -700,10 +700,14 @@ class AIAgent:
except Exception:
pass
# Direct OpenAI sessions use the Responses API path. GPT-5.x tool
# calls with reasoning are rejected on /v1/chat/completions, and
# Hermes is a tool-using client by default.
if self.api_mode == "chat_completions" and self._is_direct_openai_url():
# GPT-5.x models require the Responses API path — they are rejected
# on /v1/chat/completions by both OpenAI and OpenRouter. Also
# auto-upgrade for direct OpenAI URLs (api.openai.com) since all
# newer tool-calling models prefer Responses there.
if self.api_mode == "chat_completions" and (
self._is_direct_openai_url()
or self._model_requires_responses_api(self.model)
):
self.api_mode = "codex_responses"
# Pre-warm OpenRouter model metadata cache in a background thread.
@ -1702,6 +1706,21 @@ class AIAgent:
"""Return True when the base URL targets OpenRouter."""
return "openrouter" in self._base_url_lower
@staticmethod
def _model_requires_responses_api(model: str) -> bool:
"""Return True for models that require the Responses API path.
GPT-5.x models are rejected on /v1/chat/completions by both
OpenAI and OpenRouter (error: ``unsupported_api_for_model``).
Detect these so the correct api_mode is set regardless of
which provider is serving the model.
"""
m = model.lower()
# Strip vendor prefix (e.g. "openai/gpt-5.4" → "gpt-5.4")
if "/" in m:
m = m.rsplit("/", 1)[-1]
return m.startswith("gpt-5")
def _max_tokens_param(self, value: int) -> dict:
"""Return the correct max tokens kwarg for the current provider.
@ -5251,7 +5270,7 @@ class AIAgent:
except Exception:
pass
# Determine api_mode from provider / base URL
# Determine api_mode from provider / base URL / model
fb_api_mode = "chat_completions"
fb_base_url = str(fb_client.base_url)
if fb_provider == "openai-codex":
@ -5260,6 +5279,10 @@ class AIAgent:
fb_api_mode = "anthropic_messages"
elif self._is_direct_openai_url(fb_base_url):
fb_api_mode = "codex_responses"
elif self._model_requires_responses_api(fb_model):
# GPT-5.x models need Responses API on every provider
# (OpenRouter, Copilot, direct OpenAI, etc.)
fb_api_mode = "codex_responses"
old_model = self.model
self.model = fb_model
@ -5348,8 +5371,8 @@ class AIAgent:
to the fallback provider for every subsequent turn. Calling this at
the top of ``run_conversation()`` makes fallback turn-scoped.
The gateway creates a fresh agent per message so this is a no-op
there (``_fallback_activated`` is always False at turn start).
The gateway caches agents across messages (``_agent_cache`` in
``gateway/run.py``), so this restoration IS needed there too.
"""
if not self._fallback_activated:
return False

10
tests/gateway/test_matrix.py
View file

@ -1043,20 +1043,28 @@ class TestMatrixSyncLoop:
call_count += 1
if call_count >= 1:
adapter._closing = True
return {"rooms": {"join": {"!room:example.org": {}}}}
return {"rooms": {"join": {"!room:example.org": {}}}, "next_batch": "s1234"}
mock_crypto = MagicMock()
mock_crypto.share_keys = AsyncMock()
mock_sync_store = MagicMock()
mock_sync_store.get_next_batch = AsyncMock(return_value=None)
mock_sync_store.put_next_batch = AsyncMock()
fake_client = MagicMock()
fake_client.sync = AsyncMock(side_effect=_sync_once)
fake_client.crypto = mock_crypto
fake_client.sync_store = mock_sync_store
fake_client.handle_sync = MagicMock(return_value=[])
adapter._client = fake_client
await adapter._sync_loop()
fake_client.sync.assert_awaited_once()
mock_crypto.share_keys.assert_awaited_once()
fake_client.handle_sync.assert_called_once()
mock_sync_store.put_next_batch.assert_awaited_once_with("s1234")
class TestMatrixEncryptedSendFallback:

8
tests/run_agent/test_run_agent_codex_responses.py
View file

@ -222,6 +222,12 @@ def test_api_mode_normalizes_provider_case(monkeypatch):
def test_api_mode_respects_explicit_openrouter_provider_over_codex_url(monkeypatch):
"""GPT-5.x models need codex_responses even on OpenRouter.
OpenRouter rejects GPT-5 models on /v1/chat/completions with
``unsupported_api_for_model``. The model-level check overrides
the provider default.
"""
_patch_agent_bootstrap(monkeypatch)
agent = run_agent.AIAgent(
model="gpt-5-codex",
@ -233,7 +239,7 @@ def test_api_mode_respects_explicit_openrouter_provider_over_codex_url(monkeypat
skip_context_files=True,
skip_memory=True,
)
assert agent.api_mode == "chat_completions"
assert agent.api_mode == "codex_responses"
assert agent.provider == "openrouter"

20
website/docs/user-guide/messaging/feishu.md
View file

@ -212,7 +212,24 @@ When users click buttons or interact with interactive cards sent by the bot, the
Card action events are dispatched with `MessageType.COMMAND`, so they flow through the normal command processing pipeline.
To use this feature, enable the **Interactive Card** event in your Feishu app's event subscriptions (`card.action.trigger`).
This is also how **command approval** works — when the agent needs to run a dangerous command, it sends an interactive card with Allow Once / Session / Always / Deny buttons. The user clicks a button, and the card action callback delivers the approval decision back to the agent.
### Required Feishu App Configuration
Interactive cards require **three** configuration steps in the Feishu Developer Console. Missing any of them causes error **200340** when users click card buttons.
1. **Subscribe to the card action event:**
In **Event Subscriptions**, add `card.action.trigger` to your subscribed events.
2. **Enable the Interactive Card capability:**
In **App Features > Bot**, ensure the **Interactive Card** toggle is enabled. This tells Feishu that your app can receive card action callbacks.
3. **Configure the Card Request URL (webhook mode only):**
In **App Features > Bot > Message Card Request URL**, set the URL to the same endpoint as your event webhook (e.g. `https://your-server:8765/feishu/webhook`). In WebSocket mode this is handled automatically by the SDK.
:::warning
Without all three steps, Feishu will successfully *send* interactive cards (sending only requires `im:message:send` permission), but clicking any button will return error 200340. The card appears to work — the error only surfaces when a user interacts with it.
:::
## Media Support
@ -412,6 +429,7 @@ WebSocket and per-group ACL settings are configured via `config.yaml` under `pla
| Post messages show as plain text | The Feishu API rejected the post payload; this is normal fallback behavior. Check logs for details. |
| Images/files not received by bot | Grant `im:message` and `im:resource` permission scopes to your Feishu app |
| Bot identity not auto-detected | Grant `admin:app.info:readonly` scope, or set `FEISHU_BOT_OPEN_ID` / `FEISHU_BOT_NAME` manually |
| Error 200340 when clicking approval buttons | Enable **Interactive Card** capability and configure **Card Request URL** in the Feishu Developer Console. See [Required Feishu App Configuration](#required-feishu-app-configuration) above. |
| `Webhook rate limit exceeded` | More than 120 requests/minute from the same IP. This is usually a misconfiguration or loop. |
## Toolset