feat: add responses mode flag

Author: anvztor

README.md

@@ -31,6 +31,8 @@ Minimal `.env`:
 ```bash
 BUB_MODEL=openrouter:qwen/qwen3-coder-next
 OPENROUTER_API_KEY=your_key_here
+# Optional: use OpenAI Responses API instead of Chat Completions
+# BUB_USE_RESPONSES=true
 ```
 
 Start interactive CLI:

env.example

@@ -28,6 +28,9 @@ LLM_API_KEY=sk-...
 # Optional custom API base
 # BUB_API_BASE=https://openrouter.ai/api/v1
 
+# Use OpenAI Responses API (opt-in; keeps completions as default)
+# BUB_USE_RESPONSES=true
+
 # ---------------------------------------------------------------------------
 # Optional Ollama web tools override
 # ---------------------------------------------------------------------------

src/bub/config/settings.py

@@ -24,6 +24,7 @@ class Settings(BaseSettings):
     model: str = "openrouter:qwen/qwen3-coder-next"
     api_key: str | None = None
     api_base: str | None = None
+    use_responses: bool = False
     ollama_api_key: str | None = None
     ollama_api_base: str | None = None
     llm_api_key: str | None = Field(default=None, validation_alias="LLM_API_KEY")

src/bub/integrations/republic_client.py

@@ -30,6 +30,7 @@ def build_llm(settings: Settings, store: FileTapeStore) -> LLM:
         settings.model,
         api_key=settings.resolved_api_key,
         api_base=settings.api_base,
+        use_responses=settings.use_responses,
         tape_store=store,
         context=default_tape_context(),
         client_args=client_args,