fix(rag): lift chat output cap to 32768 tokens — long answers no longer cut off

app/config.py

@@ -51,6 +51,10 @@ class Settings(BaseSettings):
     # AND no candidate chunk matches the question lexically (see A8).
     relevance_threshold: float = 0.62
     max_context_chars: int = 24_000
+    #: Maximum output tokens a chat answer may use (owner instruction
+    #: 2026-08-22: answers must run to their natural end — the old hard
+    #: 700-token cap cut long answers off mid-sentence).
+    max_output_tokens: int = 32_768
     chunk_target_chars: int = 2_000
     chunk_overlap_chars: int = 200
     embed_batch_size: int = 16

app/rag/llm.py

@@ -175,6 +175,10 @@ class LLMClient:
         non-empty ``delta.content`` pieces. Any failure (network, HTTP,
         malformed stream) surfaces as :class:`LLMError` so the API layer can
         turn it into an SSE ``error`` event instead of a hung request.
+
+        Answers are allowed up to ``BOR_MAX_OUTPUT_TOKENS`` (default 32 768)
+        output tokens — the old hard 700-token cap cut long answers off
+        mid-sentence (owner report 2026-08-22).
         """
         try:
             # ``{role, content}`` dicts are exactly what the message params
@@ -183,7 +187,7 @@ class LLMClient:
                 model=self.settings.llm_chat_model,
                 messages=cast("list[ChatCompletionMessageParam]", messages),
                 temperature=0.4,
-                max_tokens=700,
+                max_tokens=self.settings.max_output_tokens,
                 stream=True,
             )
             async for chunk in stream: