Change: Require --ctf flag to run in enigma mode

docs/installation/docker.md

@@ -32,6 +32,7 @@ docker run --rm -it -v /var/run/docker.sock:/var/run/docker.sock \
   -v $(pwd)/keys.cfg:/app/keys.cfg \
   sweagent/swe-agent-run:latest \
   python run.py --image_name=sweagent/enigma:latest \
+  --ctf \
   --model_name gpt4 \
   --data_path /path/to/challenge.json \
   --repo_path /path/to/repo \

docs/usage/enigma.md

@@ -29,6 +29,7 @@ We will run the following command,
 ```bash
 python run.py \
   --model_name gpt4 \
+  --ctf \
   --image_name sweagent/enigma:latest \
   --data_path ../LLM_CTF_Database/2018/CSAW-Finals/misc/leaked_flag/challenge.json \
   --repo_path ../LLM_CTF_Database/2018/CSAW-Finals/misc/leaked_flag \

run.py

@@ -108,6 +108,8 @@ class ScriptArguments(FlattenedAccess, FrozenSerializable):
     raise_exceptions: bool = False
     # Dump the entire config to the log
     print_config: bool = True
+    # Run the agent in CTF mode (SWE-agent: EnIGMA)
+    ctf: bool = False
 
     @property
     def run_name(self) -> str:
@@ -515,6 +517,7 @@ def get_args(args=None) -> ScriptArguments:
             config_file=CONFIG_DIR / "default.yaml",
         ),
         actions=ActionsArguments(open_pr=False, skip_if_commits_reference_issue=True),
+        ctf=False,
     )
 
     # Nicer yaml dumping of multiline strings

tests/test_replay_ctf.py

@@ -40,6 +40,7 @@ def test_ctf_traj_replay(test_ctf_trajectories_path, traj_rel_path, ctf_data_pat
         "config/default_ctf.yaml",
         "--raise_exceptions",
         "--noprint_config",
+        "--ctf",
     ]
     args, remaining_args = get_args(args)
     main(**vars(args), forward_args=remaining_args)