mirror of
https://github.com/Alishahryar1/free-claude-code.git
synced 2026-10-20 10:59:44 +00:00
261b12f964
## Why Running local servers are omitted from startup discovery unless a local model is already selected. This leaves installed models out of Admin and connected clients, even when the server URL is configured. Fixes #1296. ## How Include LM Studio, llama.cpp and Ollama in the existing startup and Refresh models discovery pass. Bound each local model-list operation to five seconds, including admission waits and retries, and publish successful results through the shared catalog. Provider construction retains its existing generation ownership outside that deadline. Show unused default endpoints that cannot be reached as neutral availability information. Preserve actionable errors for explicit configuration, selected providers and manual Test, including a manual Test that joins background discovery. Keep DNS, TLS, HTTP and malformed-response failures visible when retries meet the deadline. Invalidate cached local models when the endpoint or proxy changes, retain inventory on transient failures at the same endpoint, and keep endpoint diagnostics visible beside discovery results. Include unavailable local providers in `fcc-doctor` reports. Shorten the README uninstall introduction to the action the user needs to take before running the commands. |
||
|---|---|---|
| .. | ||
| api | ||
| application | ||
| cli | ||
| config | ||
| contracts | ||
| core | ||
| harnesses | ||
| messaging | ||
| providers | ||
| runtime | ||
| scripts | ||
| __init__.py | ||
| code_sessions_support.py | ||
| conftest.py | ||
| provider_request_mocks.py | ||
| test_browser_artifacts.py | ||
| test_browser_server_shutdown.py | ||
| test_thread_diagnostics.py | ||
| web_tools_support.py | ||