llm-proxy-cli 0.5.2__tar.gz → 0.5.3__tar.gz

{llm_proxy_cli-0.5.2 → llm_proxy_cli-0.5.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: llm-proxy-cli
-Version: 0.5.2
+Version: 0.5.3
 Summary: A lightweight CLI tool for delegating LLM tasks to expert models across multiple providers.
 Author-email: Kerem Barbaros Karnabat <kbarbaros@hotmail.com>
 Classifier: Programming Language :: Python :: 3
@@ -29,7 +29,7 @@ Dynamic: license-file
 - **Active Liveness Verification**: Pings candidates with a minimal chat request to drop fake/gated models before they crash your task.
 - **Automatic Fallbacks**: Provide a comma-separated list of models. If one fails, it instantly falls back to the next.
 - **Circuit Breaker**: Built-in health tracking and cooldowns to prevent spamming dead endpoints.
-- **Reasoning Extraction**: Automatically extracts and formats hidden `<thought>` or `reasoning` blocks (e.g., from Nemotron).
+- **Reasoning Extraction**: Automatically extracts `reasoning_content` from natively supported models (e.g., DeepSeek-R1 or Nemotron) and outputs them to stderr.
 - **Streaming Native**: Built on the official OpenAI SDK for fast and reliable streaming chunks.
 
 ## 🏗️ Architecture & Under the Hood

{llm_proxy_cli-0.5.2 → llm_proxy_cli-0.5.3}/README.md

@@ -13,7 +13,7 @@
 - **Active Liveness Verification**: Pings candidates with a minimal chat request to drop fake/gated models before they crash your task.
 - **Automatic Fallbacks**: Provide a comma-separated list of models. If one fails, it instantly falls back to the next.
 - **Circuit Breaker**: Built-in health tracking and cooldowns to prevent spamming dead endpoints.
-- **Reasoning Extraction**: Automatically extracts and formats hidden `<thought>` or `reasoning` blocks (e.g., from Nemotron).
+- **Reasoning Extraction**: Automatically extracts `reasoning_content` from natively supported models (e.g., DeepSeek-R1 or Nemotron) and outputs them to stderr.
 - **Streaming Native**: Built on the official OpenAI SDK for fast and reliable streaming chunks.
 
 ## 🏗️ Architecture & Under the Hood

{llm_proxy_cli-0.5.2 → llm_proxy_cli-0.5.3}/llm_proxy_cli.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: llm-proxy-cli
-Version: 0.5.2
+Version: 0.5.3
 Summary: A lightweight CLI tool for delegating LLM tasks to expert models across multiple providers.
 Author-email: Kerem Barbaros Karnabat <kbarbaros@hotmail.com>
 Classifier: Programming Language :: Python :: 3
@@ -29,7 +29,7 @@ Dynamic: license-file
 - **Active Liveness Verification**: Pings candidates with a minimal chat request to drop fake/gated models before they crash your task.
 - **Automatic Fallbacks**: Provide a comma-separated list of models. If one fails, it instantly falls back to the next.
 - **Circuit Breaker**: Built-in health tracking and cooldowns to prevent spamming dead endpoints.
-- **Reasoning Extraction**: Automatically extracts and formats hidden `<thought>` or `reasoning` blocks (e.g., from Nemotron).
+- **Reasoning Extraction**: Automatically extracts `reasoning_content` from natively supported models (e.g., DeepSeek-R1 or Nemotron) and outputs them to stderr.
 - **Streaming Native**: Built on the official OpenAI SDK for fast and reliable streaming chunks.
 
 ## 🏗️ Architecture & Under the Hood

{llm_proxy_cli-0.5.2 → llm_proxy_cli-0.5.3}/llm_proxy_cli.py

@@ -9,7 +9,7 @@ import logging
 from openai import OpenAI
 from filelock import FileLock, Timeout
 
-__version__ = "0.5.2"
+__version__ = "0.5.3"
 
 # Optional import for anthropic
 try:
@@ -43,14 +43,15 @@ AUTO_DISCOVERY_TIMEOUT = 5        # seconds - keep the "auto" resolve snappy
 
 def get_api_key(provider):
     keys_file = os.path.join(get_config_dir(), "keys.json")
+    keys = {}
     if os.path.exists(keys_file):
         try:
             with open(keys_file, 'r', encoding='utf-8') as f:
                 keys = json.load(f)
-            env_name = f"{provider.upper()}_API_KEY"
         except Exception:
             pass
-    return os.environ.get(f"{provider.upper()}_API_KEY") or (keys.get(env_name) if 'keys' in locals() else None)
+    env_name = f"{provider.upper()}_API_KEY"
+    return os.environ.get(env_name) or keys.get(env_name)
 
 
 PROVIDERS = {

{llm_proxy_cli-0.5.2 → llm_proxy_cli-0.5.3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "llm-proxy-cli"
-version = "0.5.2"
+version = "0.5.3"
 authors = [
   { name="Kerem Barbaros Karnabat", email="kbarbaros@hotmail.com" }
 ]