search challenge: triangle marker on a crate's back face (rendered in vision, in digest), report tool with distance verification, smooth turn(duration), --free/--search autonomous modes, look_at crates; verified with gpt-5.6-luna

testbed/demo.py

@@ -143,6 +143,7 @@ async def run_llm_agent(
     digest: bool | None = None,
     look_every: int = 0,
     cruise: float = 0.0,
+    autonomous: bool = False,
 ) -> dict[str, Any]:
     """Autonomous LLM run: controller + nudge/correct loop (see llm_agent)."""
     from testbed.llm_agent import LLMController, run_llm_agent_loop
@@ -317,6 +318,7 @@ async def run_demo(args: argparse.Namespace) -> dict[str, Any]:
                 digest=args.digest,
                 look_every=args.look_every,
                 cruise=args.cruise,
+                autonomous=args.free,
             )
         elif args.agent == "scripted":
             summary = await run_scripted_agent(
@@ -341,6 +343,7 @@ async def run_demo(args: argparse.Namespace) -> dict[str, Any]:
                     digest=args.digest,
                     look_every=args.look_every,
                     cruise=args.cruise,
+                    autonomous=args.free,
                 )
                 if summary.get("interacted"):
                     return summary
@@ -440,6 +443,11 @@ def main() -> int:
         default=1.0,
         help="proactive motion: glide forward (meters) while the LLM thinks (0 disables)",
     )
+    parser.add_argument(
+        "--free",
+        action="store_true",
+        help="fewer restrictions: no re-aim corrections or collision hints; the model plans freely",
+    )
     parser.add_argument(
         "--frame",
         default="demo_final_frame.png",