mini-swe-agent 1.2.0__py3-none-any.whl → 1.4.0__py3-none-any.whl

{mini_swe_agent-1.2.0.dist-info → mini_swe_agent-1.4.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mini-swe-agent
-Version: 1.2.0
+Version: 1.4.0
 Summary: Nano SWE Agent - A simple AI software engineering agent
 Author-email: Kilian Lieret <kilian.lieret@posteo.de>, "Carlos E. Jimenez" <carlosej@princeton.edu>
 License: MIT License

{mini_swe_agent-1.2.0.dist-info → mini_swe_agent-1.4.0.dist-info}/RECORD

@@ -1,28 +1,28 @@
-mini_swe_agent-1.2.0.dist-info/licenses/LICENSE.md,sha256=D3luWPkdHAe7LBsdD4vzqDAXw6Xewb3G-uczss0uh1s,1094
-minisweagent/__init__.py,sha256=O834rP05yuUp1YEVIdVhl0wEGk2r43BSKUTGg7iACJU,1787
+mini_swe_agent-1.4.0.dist-info/licenses/LICENSE.md,sha256=D3luWPkdHAe7LBsdD4vzqDAXw6Xewb3G-uczss0uh1s,1094
+minisweagent/__init__.py,sha256=3w6Aygz0dJetlfY8zRfpQ4rZecN_3XvpkNLJubSeGjA,1787
 minisweagent/__main__.py,sha256=FIyAOiw--c3FQ2g240FOM1FdL0lk_PxSpixu0pQ7WFo,194
 minisweagent/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 minisweagent/agents/__init__.py,sha256=cpjJLzg1IGxLM-tZpoMJV9S33ye13XtdBO0x7DU_Lrk,48
 minisweagent/agents/default.py,sha256=6TZUfKch6e_7m05BXmjqUjV-k068s6ldg7T6zJBbQW8,5438
 minisweagent/agents/interactive.py,sha256=_DCBabdwdIR4gAojT_TaQW2MSFtmBq997mwmiDGYdRA,7327
-minisweagent/agents/interactive_textual.py,sha256=zzNsq1OkEmrBFVK3t1dxnrE7W7xU7Vc-WN47bxZcDIk,12657
+minisweagent/agents/interactive_textual.py,sha256=-GrhQT3nh_bn4qyWFRMf6F7xEqMsRnQ-Ry5n3JJBjsc,16077
 minisweagent/config/README.md,sha256=tPruhnQDhZ8ugc1FNPKk9tVMRltmmIjdYgvHCmN-3Hs,354
 minisweagent/config/__init__.py,sha256=UfORdQID1Ek_dduZlybUsIKJjihImkSqNU5tIjpw0hk,694
 minisweagent/config/default.yaml,sha256=AGhcIq6X6n5Fs71ufO3B6CtZ4PS877tCxkPkrWR5Ylg,4497
 minisweagent/config/github_issue.yaml,sha256=evvu3AJ52tXYSdami9_B8zfazOAE2r2XXkzVmScBoKc,4539
-minisweagent/config/mini.tcss,sha256=ThSOtS6JpXxqEYGX69TLy6gPZzuijngsNLI6SjnEJLY,1821
-minisweagent/config/mini.yaml,sha256=U-mTAgnrT2mn9_VxKSnLlq26Einxq0grqiY3esTtmGg,4983
+minisweagent/config/mini.tcss,sha256=VgdZZqWElA5_nn4DJUFMpz8C7Gmi5s5XOtm7pfyM83Q,1122
+minisweagent/config/mini.yaml,sha256=WluQAx4AII9MFk3xDSzsJTosNJfgZti02niCYZWYq_A,5346
 minisweagent/config/extra/__init__.py,sha256=e1MoAlDn_wc9HnXNoncf1P-B4DQ-iRf6n7Q_txjZGRI,52
 minisweagent/config/extra/swebench.yaml,sha256=LNpTahpul6HL0HozgAAz-C6kpX3wZA7Tg8uE-ZmgrF4,7577
 minisweagent/environments/__init__.py,sha256=g5mKac1YgVOZVKvmiAiuyPSevRYpI69V4vYrbCH3gsI,54
-minisweagent/environments/docker.py,sha256=woTrx-e8h_AjW60Tc-fC6m_ZQt1KCk91viLdOXm7k6g,3751
+minisweagent/environments/docker.py,sha256=VYk7i0T0IgUF_s-N-DqYkHsBWbfgaIMpJZIIdEtetTw,3871
 minisweagent/environments/local.py,sha256=-2EV3RqZSB8WEjJE7BHLhRjocPMLpoJ3HbM8QB1WXUU,1060
 minisweagent/environments/singularity.py,sha256=j7ptRVF8GwDLd-5IjhT5j7fNxEJz9amuLTmVxotaMlI,1796
 minisweagent/environments/extra/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 minisweagent/environments/extra/swerex_docker.py,sha256=MOhhFdX1sAk_U0g-GOxohfjrldzO4YfrUnHq8qJff7c,1502
 minisweagent/models/__init__.py,sha256=J4bnvfMByTVG0cL_6p51sm8gdargXhARfbG5c0UZ8Z4,2890
 minisweagent/models/anthropic.py,sha256=D8nHvvbgzPjla0He8p0O9kaXASPWg1Sai0pHsAj_Yn8,855
-minisweagent/models/litellm_model.py,sha256=5Nzln0Iq85Syluz-wGRbv5J_E6T7Mym_0M28tS8lteU,2039
+minisweagent/models/litellm_model.py,sha256=jMWsFYPPoXgBaN6Ypa5MR6A9hOpvacnYEXZkzuAkgNc,2283
 minisweagent/models/test_models.py,sha256=oB3jmZUire5TkVT8ebUCD3jLuLhPIbcTiTqdIix85Yw,1174
 minisweagent/models/utils/cache_control.py,sha256=mG9cE56HQaUwXfoqvXoH6LcbMV_G1vlEE1aBBpikXYg,1608
 minisweagent/models/utils/key_per_thread.py,sha256=Vlxt--rapNNYCgIHrMCu1WVAkuiVIhC_awbarkbnkZQ,644
@@ -30,17 +30,17 @@ minisweagent/run/__init__.py,sha256=WIoYgHVl7iZF2YncrfV3IttupG6P5KogroKHKECka3A,
 minisweagent/run/github_issue.py,sha256=GWOkGM09jOYV93p6xIM_kKWmC1yP_d5lprafWlqoBN0,2748
 minisweagent/run/hello_world.py,sha256=erLnEwNmPFLxq3-8zyv66Vy1kIqMqQf97vISX7LrQXg,959
 minisweagent/run/inspector.py,sha256=QnY3oYzm-yq3w9Jzs112Lco2Rg84vSocAWrQRVz_1lc,7127
-minisweagent/run/mini.py,sha256=9EzTUT1cra6sHkqUTGu5oqx9Esgt_XOSGW9cxFyXLd8,4339
+minisweagent/run/mini.py,sha256=Q-B5LDFQtEoqxMC3cHpbQr8IW1qNpiTh_eaQpYHz954,4589
 minisweagent/run/mini_extra.py,sha256=ecA1PnTWElpO60G9RktvVLtUOf3bZ_ESmnSttS6izhQ,1465
 minisweagent/run/extra/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 minisweagent/run/extra/config.py,sha256=paMHfplhKsqNmzhCmozxhXWHvBzBCUlwUWD8N7ytCPc,3277
-minisweagent/run/extra/swebench.py,sha256=x50nFj4tdXkdoad6TkK7tP8CSgf-WkrY0IdSMe8_oX0,9564
+minisweagent/run/extra/swebench.py,sha256=m5_PZI4ojkUyCxzkkMtel_vlnYmjziWrXu73yHoZGFs,9688
 minisweagent/run/extra/swebench_single.py,sha256=L3Kk4G65o3MCPLMEwGNIs77-AFf6Lfc8o1oxrbN-ZWM,1991
 minisweagent/run/extra/utils/batch_progress.py,sha256=u__khJ-fipZLxTJu43LamGAtPUCqEZYEi8J7SfH7X6A,6211
 minisweagent/run/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-minisweagent/run/utils/save.py,sha256=3_kuutw-uAGIhEoDawA3_FPeSz1vWuCWpJl80j5u7_s,893
-mini_swe_agent-1.2.0.dist-info/METADATA,sha256=4JI2AGHnBGJhAtkSFhmTBwH7-2QO0ymepR2g8TQGaAA,13459
-mini_swe_agent-1.2.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-mini_swe_agent-1.2.0.dist-info/entry_points.txt,sha256=d1_yRbTaGjs1UXHa6JQK0sKDGBIVGm8oeW0k2kfbJgQ,182
-mini_swe_agent-1.2.0.dist-info/top_level.txt,sha256=zKF4t8bFpV87fdVABZt2Da-vnb4Vkh_CxkwQx5YT4Ew,13
-mini_swe_agent-1.2.0.dist-info/RECORD,,
+minisweagent/run/utils/save.py,sha256=q7omf7zYHg73k8-Iyp9w5YVSYvDAacRrh4X9L_4VhNM,942
+mini_swe_agent-1.4.0.dist-info/METADATA,sha256=tM-cskbK7S2xIRUegGm39qXgPT3tOfpfZOlOZfk-lcg,13459
+mini_swe_agent-1.4.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+mini_swe_agent-1.4.0.dist-info/entry_points.txt,sha256=d1_yRbTaGjs1UXHa6JQK0sKDGBIVGm8oeW0k2kfbJgQ,182
+mini_swe_agent-1.4.0.dist-info/top_level.txt,sha256=zKF4t8bFpV87fdVABZt2Da-vnb4Vkh_CxkwQx5YT4Ew,13
+mini_swe_agent-1.4.0.dist-info/RECORD,,

minisweagent/__init__.py

@@ -8,7 +8,7 @@ This file provides:
   unless you want the static type checking.
 """
 
-__version__ = "1.2.0"
+__version__ = "1.4.0"
 
 import os
 from pathlib import Path

minisweagent/agents/interactive_textual.py

@@ -8,6 +8,7 @@ import os
 import re
 import threading
 import time
+import traceback
 from dataclasses import dataclass, field
 from pathlib import Path
 from typing import Literal
@@ -19,9 +20,9 @@ from textual.binding import Binding
 from textual.containers import Container, Vertical, VerticalScroll
 from textual.css.query import NoMatches
 from textual.events import Key
-from textual.widgets import Footer, Header, Static, TextArea
+from textual.widgets import Footer, Header, Input, Static, TextArea
 
-from minisweagent.agents.default import AgentConfig, DefaultAgent, NonTerminatingException
+from minisweagent.agents.default import AgentConfig, DefaultAgent, NonTerminatingException, Submitted
 
 
 @dataclass
@@ -30,6 +31,8 @@ class TextualAgentConfig(AgentConfig):
     """Mode for action execution: 'confirm' requires user confirmation, 'yolo' executes immediately."""
     whitelist_actions: list[str] = field(default_factory=list)
     """Never confirm actions that match these regular expressions."""
+    confirm_exit: bool = True
+    """If the agent wants to finish, do we ask for confirmation from user?"""
 
 
 class TextualAgent(DefaultAgent):
@@ -37,17 +40,29 @@ class TextualAgent(DefaultAgent):
         """Connects the DefaultAgent to the TextualApp."""
         self.app = app
         super().__init__(*args, config_class=TextualAgentConfig, **kwargs)
+        self._current_action_from_human = False
 
     def add_message(self, role: str, content: str):
         super().add_message(role, content)
         if self.app.agent_state != "UNINITIALIZED":
             self.app.call_from_thread(self.app.on_message_added)
 
+    def query(self) -> dict:
+        if self.config.mode == "human":
+            human_input = self.app.input_container.request_input("Enter your command:")
+            self._current_action_from_human = True
+            msg = {"content": f"\n```bash\n{human_input}\n```"}
+            self.add_message("assistant", msg["content"])
+            return msg
+        self._current_action_from_human = False
+        return super().query()
+
     def run(self, task: str) -> tuple[str, str]:
         try:
             exit_status, result = super().run(task)
         except Exception as e:
             result = str(e)
+            print(traceback.format_exc())
             self.app.call_from_thread(self.app.on_agent_finished, "ERROR", result)
             return "ERROR", result
         else:
@@ -55,13 +70,30 @@ class TextualAgent(DefaultAgent):
         return exit_status, result
 
     def execute_action(self, action: dict) -> dict:
-        if self.config.mode == "confirm" and not any(
-            re.match(r, action["action"]) for r in self.config.whitelist_actions
+        if self.config.mode == "human" and not self._current_action_from_human:  # threading, grrrrr
+            raise NonTerminatingException("Command not executed because user switched to manual mode.")
+        if (
+            self.config.mode == "confirm"
+            and action["action"].strip()
+            and not any(re.match(r, action["action"]) for r in self.config.whitelist_actions)
         ):
-            if result := self.app.confirmation_container.request_confirmation(action["action"]):
+            result = self.app.input_container.request_input("Press ENTER to confirm or provide rejection reason")
+            if result:  # Non-empty string means rejection
                 raise NonTerminatingException(f"Command not executed: {result}")
         return super().execute_action(action)
 
+    def has_finished(self, output: dict[str, str]):
+        try:
+            return super().has_finished(output)
+        except Submitted as e:
+            if self.config.confirm_exit:
+                if new_task := self.app.input_container.request_input(
+                    "[bold green]Agent wants to finish.[/bold green] "
+                    "[green]Type a comment to give it a new task or press enter to quit.\n"
+                ).strip():
+                    raise NonTerminatingException(f"The user added a new task: {new_task}")
+            raise e
+
 
 class AddLogEmitCallback(logging.Handler):
     def __init__(self, callback):
@@ -87,77 +119,123 @@ def _messages_to_steps(messages: list[dict]) -> list[list[dict]]:
     return steps
 
 
-class ConfirmationPromptContainer(Container):
+class SmartInputContainer(Container):
     def __init__(self, app: "AgentApp"):
-        """This class is responsible for handling the action execution confirmation."""
-        super().__init__(id="confirmation-container")
+        """Smart input container supporting single-line and multi-line input modes."""
+        super().__init__(classes="smart-input-container")
         self._app = app
-        self.rejecting = False
+        self._multiline_mode = False
         self.can_focus = True
         self.display = False
 
-        self._pending_action: str | None = None
-        self._confirmation_event = threading.Event()
-        self._confirmation_result: str | None = None
+        self.pending_prompt: str | None = None
+        self._input_event = threading.Event()
+        self._input_result: str | None = None
 
-    def compose(self) -> ComposeResult:
-        yield Static(
-            "Press [bold]ENTER[/bold] to confirm action or [bold]BACKSPACE[/bold] to reject (or [bold]y[/bold] to toggle YOLO mode)",
-            classes="confirmation-prompt",
+        self._header_display = Static(
+            "USER INPUT REQUESTED", id="input-header-display", classes="message-header input-request-header"
         )
-        yield TextArea(id="rejection-input")
-        rejection_help = Static(
-            "Press [bold]Ctrl+D[/bold] to submit rejection message",
-            id="rejection-help",
-            classes="rejection-help",
+        self._hint_text = Static(
+            "[bold]Enter[/bold] to submit, [bold]Ctrl+T[/bold] to switch to multi-line input, [bold]Tab[/bold] to switch focus with other controls",
+            classes="hint-text",
+        )
+        self._single_input = Input(placeholder="Type your input...")
+        self._multi_input = TextArea("", show_line_numbers=False, classes="multi-input")
+
+        self._input_elements_container = Vertical(
+            self._header_display,
+            self._hint_text,
+            self._single_input,
+            self._multi_input,
+            classes="message-container",
         )
-        rejection_help.display = False
-        yield rejection_help
-
-    def request_confirmation(self, action: str) -> str | None:
-        """Request confirmation for an action. Returns rejection message or None."""
-        self._confirmation_event.clear()
-        self._confirmation_result = None
-        self._pending_action = action
-        self._app.call_from_thread(self._app.update_content)
-        self._confirmation_event.wait()
-        return self._confirmation_result
 
-    def _complete_confirmation(self, rejection_message: str | None):
-        """Internal method to complete the confirmation process."""
-        self._confirmation_result = rejection_message
-        self._pending_action = None
+    def compose(self) -> ComposeResult:
+        yield self._input_elements_container
+
+    def on_mount(self) -> None:
+        """Initialize the widget state."""
+        self._multi_input.display = False
+        self._update_mode_display()
+
+    def on_focus(self) -> None:
+        """Called when the container gains focus."""
+        if self._multiline_mode:
+            self._multi_input.focus()
+        else:
+            self._single_input.focus()
+
+    def request_input(self, prompt: str) -> str:
+        """Request input from user. Returns input text (empty string if confirmed without reason)."""
+        self._input_event.clear()
+        self._input_result = None
+        self.pending_prompt = prompt
+        self._header_display.update(prompt)
+        self._update_mode_display()
+        self._app.call_from_thread(self._app.update_content)
+        self._input_event.wait()
+        return self._input_result or ""
+
+    def _complete_input(self, input_text: str):
+        """Internal method to complete the input process."""
+        self._input_result = input_text
+        self.pending_prompt = None
+        self._header_display.update("USER INPUT REQUESTED")
         self.display = False
-        self.rejecting = False
-        rejection_input = self.query_one("#rejection-input", TextArea)
-        rejection_input.display = False
-        rejection_input.text = ""
-        rejection_help = self.query_one("#rejection-help", Static)
-        rejection_help.display = False
-        # Reset agent state to RUNNING after confirmation is completed
-        if rejection_message is None:
-            self._app.agent_state = "RUNNING"
-        self._confirmation_event.set()
+        self._single_input.value = ""
+        self._multi_input.text = ""
+        self._multiline_mode = False
+        self._update_mode_display()
+        self._app.agent_state = "RUNNING"
         self._app.update_content()
+        # Reset scroll position to bottom since input container disappearing changes layout
+        # somehow scroll_to doesn't work.
+        self._app._vscroll.scroll_y = 0
+        self._input_event.set()
+
+    def action_toggle_mode(self) -> None:
+        """Switch from single-line to multi-line mode (one-way only)."""
+        if self.pending_prompt is None or self._multiline_mode:
+            return
+
+        self._multiline_mode = True
+        self._update_mode_display()
+        self.on_focus()
+
+    def _update_mode_display(self) -> None:
+        """Update the display based on current mode."""
+        if self._multiline_mode:
+            self._multi_input.text = self._single_input.value
+            self._single_input.display = False
+            self._multi_input.display = True
+
+        else:
+            self._multi_input.display = False
+            self._single_input.display = True
+
+    def on_input_submitted(self, event: Input.Submitted) -> None:
+        """Handle single-line input submission."""
+        if not self._multiline_mode:
+            text = event.input.value.strip()
+            self._complete_input(text)
 
     def on_key(self, event: Key) -> None:
-        if self.rejecting and event.key == "ctrl+d":
+        """Handle key events."""
+        if event.key == "ctrl+t" and not self._multiline_mode:
             event.prevent_default()
-            rejection_input = self.query_one("#rejection-input", TextArea)
-            self._complete_confirmation(rejection_input.text)
+            self.action_toggle_mode()
+            return
+
+        if self._multiline_mode and event.key == "ctrl+d":
+            event.prevent_default()
+            self._complete_input(self._multi_input.text.strip())
+            return
+
+        if event.key == "escape":
+            event.prevent_default()
+            self.can_focus = False
+            self._app.set_focus(None)
             return
-        if not self.rejecting:
-            if event.key == "enter":
-                event.prevent_default()
-                self._complete_confirmation(None)
-            elif event.key == "backspace":
-                event.prevent_default()
-                self.rejecting = True
-                rejection_input = self.query_one("#rejection-input", TextArea)
-                rejection_input.display = True
-                rejection_input.focus()
-                rejection_help = self.query_one("#rejection-help", Static)
-                rejection_help.display = True
 
 
 class AgentApp(App):
@@ -171,6 +249,7 @@ class AgentApp(App):
         Binding("q", "quit", "Quit"),
         Binding("y", "yolo", "Switch to YOLO Mode"),
         Binding("c", "confirm", "Switch to Confirm Mode"),
+        Binding("u", "human", "Switch to Human Mode"),
     ]
 
     def __init__(self, model, env, task: str, **kwargs):
@@ -182,13 +261,15 @@ class AgentApp(App):
         self.agent = TextualAgent(self, model=model, env=env, **kwargs)
         self._i_step = 0
         self.n_steps = 1
-        self.confirmation_container = ConfirmationPromptContainer(self)
+        self.input_container = SmartInputContainer(self)
         self.log_handler = AddLogEmitCallback(lambda record: self.call_from_thread(self.on_log_message_emitted, record))
         logging.getLogger().addHandler(self.log_handler)
         self._spinner = Spinner("dots")
         self.exit_status: str | None = None
         self.result: str | None = None
 
+        self._vscroll = VerticalScroll()
+
     # --- Basics ---
 
     @property
@@ -201,15 +282,16 @@ class AgentApp(App):
         """Set current step index, automatically clamping to valid bounds."""
         if value != self._i_step:
             self._i_step = max(0, min(value, self.n_steps - 1))
-            self.query_one(VerticalScroll).scroll_to(y=0, animate=False)
+            self._vscroll.scroll_to(y=0, animate=False)
             self.update_content()
 
     def compose(self) -> ComposeResult:
         yield Header()
         with Container(id="main"):
-            with VerticalScroll():
-                yield Vertical(id="content")
-            yield self.confirmation_container
+            with self._vscroll:
+                with Vertical(id="content"):
+                    pass
+                yield self.input_container
         yield Footer()
 
     def on_mount(self) -> None:
@@ -221,8 +303,7 @@ class AgentApp(App):
     # --- Reacting to events ---
 
     def on_message_added(self) -> None:
-        vs = self.query_one(VerticalScroll)
-        auto_follow = self.i_step == self.n_steps - 1 and vs.scroll_target_y <= 1
+        auto_follow = self.i_step == self.n_steps - 1 and self._vscroll.scroll_y <= 1
         self.n_steps = len(_messages_to_steps(self.agent.messages))
         self.update_content()
         if auto_follow:
@@ -267,13 +348,11 @@ class AgentApp(App):
             message_container.mount(Static(role.upper(), classes="message-header"))
             message_container.mount(Static(Text(content_str, no_wrap=False), classes="message-content"))
 
-        if self.confirmation_container._pending_action is not None:
-            self.agent_state = "AWAITING_CONFIRMATION"
-        self.confirmation_container.display = (
-            self.confirmation_container._pending_action is not None and self.i_step == len(items) - 1
-        )
-        if self.confirmation_container.display:
-            self.confirmation_container.focus()
+        if self.input_container.pending_prompt is not None:
+            self.agent_state = "AWAITING_INPUT"
+        self.input_container.display = self.input_container.pending_prompt is not None and self.i_step == len(items) - 1
+        if self.input_container.display:
+            self.input_container.on_focus()
 
         self._update_headers()
         self.refresh()
@@ -294,12 +373,21 @@ class AgentApp(App):
 
     def action_yolo(self):
         self.agent.config.mode = "yolo"
-        self.confirmation_container._complete_confirmation(None)
-        self.notify("YOLO mode enabled - actions will execute immediately")
+        if self.input_container.pending_prompt is not None:
+            self.input_container._complete_input("")  # accept
+        self.notify("YOLO mode enabled - LM actions will execute immediately")
+
+    def action_human(self):
+        if self.agent.config.mode == "confirm" and self.input_container.pending_prompt is not None:
+            self.input_container._complete_input("User switched to manual mode, this command will be ignored")
+        self.agent.config.mode = "human"
+        self.notify("Human mode enabled - you can now type commands directly")
 
     def action_confirm(self):
+        if self.agent.config.mode == "human" and self.input_container.pending_prompt is not None:
+            self.input_container._complete_input("")  # just submit blank action
         self.agent.config.mode = "confirm"
-        self.notify("Confirm mode enabled - actions will require confirmation")
+        self.notify("Confirm mode enabled - LM proposes commands and you confirm/reject them")
 
     def action_next_step(self) -> None:
         self.i_step += 1
@@ -314,9 +402,7 @@ class AgentApp(App):
         self.i_step = self.n_steps - 1
 
     def action_scroll_down(self) -> None:
-        vs = self.query_one(VerticalScroll)
-        vs.scroll_to(y=vs.scroll_target_y + 15)
+        self._vscroll.scroll_to(y=self._vscroll.scroll_target_y + 15)
 
     def action_scroll_up(self) -> None:
-        vs = self.query_one(VerticalScroll)
-        vs.scroll_to(y=vs.scroll_target_y - 15)
+        self._vscroll.scroll_to(y=self._vscroll.scroll_target_y - 15)

minisweagent/config/mini.tcss

@@ -1,12 +1,13 @@
 Screen {
     layout: grid;
     grid-size: 1;
-    grid-rows: 1fr 8 1fr;
+    grid-rows: auto 1fr auto;
 }
 
 #main {
     height: 100%;
     padding: 1;
+    layout: vertical;
 }
 
 Footer {
@@ -16,6 +17,32 @@ Footer {
 
 #content {
     height: auto;
+    min-height: 0;
+}
+
+.smart-input-container {
+    height: auto;
+    margin-top: 0;
+    padding: 0;
+    min-height: 0;
+}
+
+.multi-input {
+    height: auto;
+    max-height: 20;
+    min-height: 3;
+}
+
+.prompt-display {
+    margin-bottom: 1;
+    padding: 0 1;
+    text-style: bold;
+}
+
+.hint-text{
+    margin-bottom: 1;
+    padding: 0 1;
+    color: white;
 }
 
 .message-container {
@@ -33,6 +60,10 @@ Footer {
     text-style: bold;
 }
 
+.input-request-header {
+    color: $warning;
+}
+
 .message-content {
     margin-top: 1;
     padding: 0 1;
@@ -42,29 +73,6 @@ Header.running {
     background: $error;
 }
 
-.confirmation-modal {
-    layout: vertical;
-    background: $surface;
-    margin: 1 4;
-    min-width: 40;
-    padding: 1;
-    border: tall $primary;
-    height: auto;
-}
-
-.modal-title {
-    text-align: center;
-    text-style: bold;
-    margin-bottom: 1;
-}
-
-.modal-content {
-    margin: 1 2;
-    min-height: 1;
-    max-height: 10;
-    overflow-y: auto;
-}
-
 .button-container {
     layout: horizontal;
     align-horizontal: center;
@@ -74,55 +82,4 @@ Header.running {
 .button-container Button {
     margin: 0 1;
     min-width: 10;
-}
-
-.confirmation-container {
-    background: $boost;
-    border: heavy $primary;
-    padding: 1;
-    margin: 1;
-}
-
-.confirmation-header {
-    color: $warning;
-    text-style: bold;
-}
-
-.command-to-confirm {
-    background: $surface;
-    margin: 1 0;
-    padding: 1;
-    color: $text;
-}
-
-#confirmation-input {
-    margin-top: 1;
-}
-
-.confirmation-prompt {
-    background: $boost;
-    border: heavy $warning;
-    padding: 1;
-    margin: 1;
-    text-align: center;
-    color: $warning;
-}
-
-.confirmation-prompt:focus {
-    border: heavy $accent;
-    background: $panel;
-}
-
-#rejection-input {
-    display: none;
-    margin: 1;
-}
-
-.rejection-help {
-    background: $boost;
-    border: heavy $warning;
-    padding: 1;
-    margin: 1;
-    text-align: center;
-    color: $warning;
 }

minisweagent/config/mini.yaml

@@ -23,13 +23,15 @@ agent:
     You can execute bash commands and edit files to implement the necessary changes.
 
     ## Recommended Workflow
-    1. Analyze the codebase by finding and reading relevant files.
-       If present, you might want to take a look at the following files that set additional guidelines
-       for your work: CLAUDE.md, .cursor/rules/<relevant rules>
-    2. Create a script to reproduce the issue
-    3. Edit the source code to resolve the issue
-    4. Verify your fix works by running your script again
-    5. Test edge cases to ensure your fix is robust
+
+    1. If present, you might want to take a look at the following files that set additional guidelines
+       for your work: CLAUDE.md, .github/copilot-instructions.md. Use a find command to locate all of them at once.
+       If files corresponding to multiple AI systems are present, it's enough to read the ones for one of them.
+    2. Analyze the codebase by finding and reading relevant files.
+    3. Create a script to reproduce the issue
+    4. Edit the source code to resolve the issue
+    5. Verify your fix works by running your script again
+    6. Test edge cases to ensure your fix is robust
 
     ## Important Rules
 
@@ -140,6 +142,9 @@ agent:
     <action>
     ```
     </response_example>
+
+    Note: In rare cases, if you need to reference a similar format in your command, you might have
+    to proceed in two steps, first writing TRIPLEBACKTICKSBASH, then replacing them with ```bash.
   step_limit: 0.
   cost_limit: 3.
   mode: confirm

minisweagent/environments/docker.py

@@ -24,6 +24,8 @@ class DockerEnvironmentConfig:
     """Path to the docker/container executable."""
     run_args: list[str] = field(default_factory=list)
     """Additional arguments to pass to the docker/container executable."""
+    container_timeout: str = "2h"
+    """Max duration to keep container running. Uses the same format as the sleep command."""
 
 
 class DockerEnvironment:
@@ -49,7 +51,7 @@ class DockerEnvironment:
             *self.config.run_args,
             self.config.image,
             "sleep",
-            "infinity",  # Keep container running
+            self.config.container_timeout,
         ]
         print(f"Starting container with command: {shlex.join(cmd)}")
         result = subprocess.run(

minisweagent/models/litellm_model.py

@@ -1,5 +1,7 @@
+import json
 import logging
 from dataclasses import dataclass, field
+from pathlib import Path
 from typing import Any
 
 import litellm
@@ -20,6 +22,7 @@ logger = logging.getLogger("litellm_model")
 class LitellmModelConfig:
     model_name: str
     model_kwargs: dict[str, Any] = field(default_factory=dict)
+    litellm_model_registry: Path | None = None
 
 
 class LitellmModel:
@@ -27,6 +30,8 @@ class LitellmModel:
         self.config = LitellmModelConfig(**kwargs)
         self.cost = 0.0
         self.n_calls = 0
+        if self.config.litellm_model_registry is not None:
+            litellm.utils.register_model(json.loads(self.config.litellm_model_registry.read_text()))
 
     @retry(
         stop=stop_after_attempt(10),

minisweagent/run/extra/swebench.py

@@ -1,11 +1,7 @@
 #!/usr/bin/env python3
 
-"""Run mini-SWE-agent on SWEBench instances.
-
-[not dim]
-More information about the usage: [bold green]https://mini-swe-agent.com/latest/usage/swebench/[/bold green]
-[/not dim]
-"""
+"""Run mini-SWE-agent on SWE-bench instances in batch mode."""
+# Read this first: https://mini-swe-agent.com/latest/usage/swebench/  (usage docs)
 
 import concurrent.futures
 import json
@@ -28,6 +24,13 @@ from minisweagent.models import get_model
 from minisweagent.run.extra.utils.batch_progress import RunBatchProgressManager
 from minisweagent.run.utils.save import save_traj
 
+_HELP_TEXT = """Run mini-SWE-agent on SWEBench instances.
+
+[not dim]
+More information about the usage: [bold green]https://mini-swe-agent.com/latest/usage/swebench/[/bold green]
+[/not dim]
+"""
+
 app = typer.Typer(rich_markup_mode="rich", add_completion=False)
 
 DATASET_MAPPING = {
@@ -168,7 +171,7 @@ def filter_instances(
     return instances
 
 
-@app.command()
+@app.command(help=_HELP_TEXT)
 def main(
     subset: str = typer.Option("lite", "--subset", help="SWEBench subset to use or path to a dataset"),
     split: str = typer.Option("dev", "--split", help="Dataset split"),
@@ -183,7 +186,6 @@ def main(
         builtin_config_dir / "extra" / "swebench.yaml", "-c", "--config", help="Path to a config file"
     ),
 ) -> None:
-    """Run mini-SWE-agent on SWEBench instances"""
     dataset_path = DATASET_MAPPING.get(subset, subset)
     print(f"Loading dataset {dataset_path}, split {split}...")
     instances = list(load_dataset(dataset_path, split=split))

minisweagent/run/mini.py

@@ -1,16 +1,7 @@
 #!/usr/bin/env python3
 
-"""Run mini-SWE-agent in your local environment.
-
-[not dim]
-There are two different user interfaces:
-
-[bold green]mini[/bold green] Simple REPL-style interface
-[bold green]mini -v[/bold green] Pager-style interface (Textual)
-
-More information about the usage: [bold green]https://mini-swe-agent.com/latest/usage/mini/[/bold green]
-[/not dim]
-"""
+"""Run mini-SWE-agent in your local environment. This is the default executable `mini`."""
+# Read this first: https://mini-swe-agent.com/latest/usage/mini/  (usage)
 
 import os
 from pathlib import Path
@@ -33,9 +24,21 @@ from minisweagent.run.extra.config import configure_if_first_time
 from minisweagent.run.utils.save import save_traj
 
 DEFAULT_CONFIG = Path(os.getenv("MSWEA_MINI_CONFIG_PATH", builtin_config_dir / "mini.yaml"))
+DEFAULT_OUTPUT = global_config_dir / "last_mini_run.traj.json"
 console = Console(highlight=False)
 app = typer.Typer(rich_markup_mode="rich")
 prompt_session = PromptSession(history=FileHistory(global_config_dir / "mini_task_history.txt"))
+_HELP_TEXT = """Run mini-SWE-agent in your local environment.
+
+[not dim]
+There are two different user interfaces:
+
+[bold green]mini[/bold green] Simple REPL-style interface
+[bold green]mini -v[/bold green] Pager-style interface (Textual)
+
+More information about the usage: [bold green]https://mini-swe-agent.com/latest/usage/mini/[/bold green]
+[/not dim]
+"""
 
 
 def run_interactive(model: Model, env: Environment, agent_config: dict, task: str, output: Path | None = None) -> Any:
@@ -68,7 +71,7 @@ def run_textual(model: Model, env: Environment, agent_config: dict, task: str, o
             save_traj(agent_app.agent, output, exit_status=agent_app.exit_status, result=agent_app.result)
 
 
-@app.command(help=__doc__)
+@app.command(help=_HELP_TEXT)
 def main(
     visual: bool = typer.Option(False, "-v", "--visual", help="Use visual (pager-style) UI (Textual)"),
     model_name: str | None = typer.Option(
@@ -81,7 +84,7 @@ def main(
     yolo: bool = typer.Option(False, "-y", "--yolo", help="Run without confirmation"),
     cost_limit: float | None = typer.Option(None, "-l", "--cost-limit", help="Cost limit. Set to 0 to disable."),
     config_spec: Path = typer.Option(DEFAULT_CONFIG, "-c", "--config", help="Path to config file"),
-    output: Path | None = typer.Option(None, "-o", "--output", help="Output file"),
+    output: Path | None = typer.Option(DEFAULT_OUTPUT, "-o", "--output", help="Output trajectory file"),
     exit_immediately: bool = typer.Option(
         False, "--exit-immediately", help="Exit immediately when the agent wants to finish instead of prompting."
     ),
@@ -105,7 +108,7 @@ def main(
     config["agent"]["mode"] = "confirm" if not yolo else "yolo"
     if cost_limit:
         config["agent"]["cost_limit"] = cost_limit
-    if not visual and exit_immediately:
+    if exit_immediately:
         config["agent"]["confirm_exit"] = False
     model = get_model(model_name, config.get("model", {}))
     env = LocalEnvironment(**config.get("env", {}))

minisweagent/run/utils/save.py

@@ -23,6 +23,7 @@ def save_traj(
             },
         },
         "messages": [],
+        "trajectory_format": "mini-swe-agent-1",
     } | kwargs
     if agent is not None:
         data["info"]["model_stats"]["instance_cost"] = agent.model.cost