Fix rotation failover: advance past unavailable providers instead of retrying one (v0.99.99)

The dispatch loop skipped a provider via is_rate_limited() (which includes the
manual dashboard flag) but did NOT add it to tried_models, so the next retry
re-selected the SAME highest-weight provider every time. A manually-flagged /
rate-limited top-weight member (e.g. claude_think in rotation lisa) thus consumed
all max_retries and returned 429, never failing over to a working member
(kilo-stefy).

- Use _provider_in_availability_cooldown() instead of is_rate_limited() so the
  manual flag (rotation-membership only) does not block dispatch, consistent with
  the scan.
- Mark the skipped model as tried so failover actually advances to the next
  provider.
- Guard the dispatch handler build in try/except too (same bad-cred 500 class).

rotation/lisa now returns 200 via kilo-stefy instead of 429/500.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

aisbf/__init__.py

@@ -55,7 +55,7 @@ from .auth.qwen import QwenOAuth2
 from .handlers import RequestHandler, RotationHandler, AutoselectHandler
 from .utils import count_messages_tokens, split_messages_into_chunks, get_max_request_tokens_for_model, get_max_completion_tokens_for_model
 
-__version__ = "0.99.98"
+__version__ = "0.99.99"
 __all__ = [
     # Config
     "config",

aisbf/handlers.py

@@ -3846,11 +3846,24 @@ class RotationHandler:
                     continue
 
             logger.info(f"Getting provider handler for {provider_id}")
-            handler = get_provider_handler(provider_id, api_key, user_id=self.user_id)
+            try:
+                handler = get_provider_handler(provider_id, api_key, user_id=self.user_id)
+            except Exception as e:
+                logger.warning(f"Provider {provider_id} handler build failed ({e}); failing over to next model")
+                tried_models.append(current_model)
+                continue
             logger.info(f"Provider handler obtained: {handler.__class__.__name__}")
 
-            if handler.is_rate_limited():
-                logger.warning(f"Provider {provider_id} is rate limited, skipping to next model")
+            # Only a genuine auto-disable cooldown (failure threshold / usage limit)
+            # skips a provider here. The manual dashboard flag is rotation-membership
+            # only (honoured by enabled:false in the scan), so it must NOT block
+            # dispatch — otherwise the scan marks a manually-flagged provider available
+            # but dispatch skips it. Crucially, mark the skipped model as tried so
+            # failover advances to the next provider instead of re-selecting this same
+            # one every retry (which exhausted max_retries and returned 429).
+            if _provider_in_availability_cooldown(handler):
+                logger.warning(f"Provider {provider_id} is in an auto-disable cooldown, failing over to next model")
+                tried_models.append(current_model)
                 continue
             
             # Check token rate limits for this model

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "aisbf"
-version = "0.99.98"
+version = "0.99.99"
 description = "AISBF - AI Service Broker Framework || AI Should Be Free - A modular proxy server for managing multiple AI provider integrations"
 readme = "README.md"
 license = "GPL-3.0-or-later"

setup.py

@@ -106,7 +106,7 @@ class InstallCommand(_install):
 
 setup(
     name="aisbf",
-    version="0.99.98",
+    version="0.99.99",
     author="AISBF Contributors",
     author_email="stefy@nexlab.net",
     description="AISBF - AI Service Broker Framework || AI Should Be Free - A modular proxy server for managing multiple AI provider integrations",