Honor interactive model override after current CLI changes (#111)

The interactive CLI accepted --model but did not pass it into main(), so the REPL used the configured default while headless mode honored the flag. Reapply the override on top of the current token/session plumbing by changing only the interactive main signature, config override, and CLI dispatch.

Constraint: Current main already passes user_id into submission_loop and uses shared HF token resolution; this fix must preserve both paths.

Rejected: Reuse the old PR diff verbatim | it drops newer user_id/session identity changes and conflicts with current main.

Confidence: high

Scope-risk: narrow

Directive: Keep --model behavior identical for interactive and headless paths; banner should render the resolved model.

Tested: python -m py_compile agent/main.py tests/unit/test_cli_rendering.py

Tested: UV_CACHE_DIR=/tmp/uv-cache uv run --extra dev pytest tests/unit/test_cli_rendering.py -q

Tested: UV_CACHE_DIR=/tmp/uv-cache uv run --extra dev pytest tests/unit/test_llm_params.py tests/unit/test_hf_access.py tests/unit/test_user_quotas.py -q

Co-authored-by: akseljoonas <aksel.reedi@gmail.com>

agent/main.py

@@ -807,7 +807,7 @@ async def _handle_slash_command(
     return None
 
 
-async def main():
+async def main(model: str | None = None):
     """Interactive chat with the agent"""
 
     # Clear screen
@@ -822,6 +822,8 @@ async def main():
         hf_token = await _prompt_and_save_hf_token(prompt_session)
 
     config = load_config(CLI_CONFIG_PATH)
+    if model:
+        config.model_name = model
 
     # Resolve username for banner
     hf_user = _get_hf_user(hf_token)
@@ -1240,7 +1242,7 @@ def cli():
                 max_iter = 10_000  # effectively unlimited
             asyncio.run(headless_main(args.prompt, model=args.model, max_iterations=max_iter, stream=not args.no_stream))
         else:
-            asyncio.run(main())
+            asyncio.run(main(model=args.model))
     except KeyboardInterrupt:
         print("\n\nGoodbye!")
 

tests/unit/test_cli_rendering.py

@@ -1,8 +1,12 @@
 """Regression tests for interactive CLI rendering and research model routing."""
 
+import sys
 from io import StringIO
 from types import SimpleNamespace
 
+import pytest
+
+import agent.main as main_mod
 from agent.tools.research_tool import _get_research_model
 from agent.utils import terminal_display
 
@@ -42,3 +46,45 @@ def test_subagent_display_does_not_spawn_background_redraw(monkeypatch):
     mgr.clear("agent-1")
 
     assert calls == []
+
+
+def test_cli_forwards_model_flag_to_interactive_main(monkeypatch):
+    seen: dict[str, str | None] = {}
+
+    async def fake_main(*, model=None):
+        seen["model"] = model
+
+    monkeypatch.setattr(sys, "argv", ["ml-intern", "--model", "openai/gpt-5.5"])
+    monkeypatch.setattr(main_mod, "main", fake_main)
+
+    main_mod.cli()
+
+    assert seen["model"] == "openai/gpt-5.5"
+
+
+@pytest.mark.asyncio
+async def test_interactive_main_applies_model_override_before_banner(monkeypatch):
+    class StopAfterBanner(Exception):
+        pass
+
+    def fake_banner(*, model=None, hf_user=None):
+        assert model == "openai/gpt-5.5"
+        assert hf_user == "tester"
+        raise StopAfterBanner
+
+    monkeypatch.setattr(main_mod.os, "system", lambda *_args, **_kwargs: 0)
+    monkeypatch.setattr(main_mod, "PromptSession", lambda: object())
+    monkeypatch.setattr(main_mod, "resolve_hf_token", lambda: "hf-token")
+    monkeypatch.setattr(main_mod, "_get_hf_user", lambda _token: "tester")
+    monkeypatch.setattr(
+        main_mod,
+        "load_config",
+        lambda _path: SimpleNamespace(
+            model_name="moonshotai/Kimi-K2.6",
+            mcpServers={},
+        ),
+    )
+    monkeypatch.setattr(main_mod, "print_banner", fake_banner)
+
+    with pytest.raises(StopAfterBanner):
+        await main_mod.main(model="openai/gpt-5.5")