Final answers cut off mid-sentence — max_tokens=300 too low for prompt's response length

src/chat_langchain_lite/agent.py

@@ -48,7 +48,7 @@ def build_agent():
         # `model_name` (and max_tokens -> max_tokens_to_sample); ty reads the alias as
         # the param name and misflags the documented kwargs, so the line below is
         # suppressed. Correct at runtime (model resolves to a.model).
-        model=ChatAnthropic(model=_model_id(), max_tokens=300, temperature=0),  # ty: ignore[unknown-argument, missing-argument]
+        model=ChatAnthropic(model=_model_id(), max_tokens=2048, temperature=0),  # ty: ignore[unknown-argument, missing-argument]
         tools=TOOLS,
         system_prompt=SYSTEM_PROMPT,
         middleware=[_readonly_context_hub_fs()],

src/chat_langchain_lite/context_hub.py

@@ -56,7 +56,7 @@
 
 - Use bullet lists wherever possible — they scan better than paragraphs
 - Wrap commands in single backticks like `pip install ...`
-- Keep responses tight (under 200 words when you can)
+- Aim for clear, complete responses; brevity is good but always finish the thought
 
 ## Respecting User Preferences