mini-swe-agent 1.4.2__tar.gz → 1.5.1__tar.gz

{mini_swe_agent-1.4.2/src/mini_swe_agent.egg-info → mini_swe_agent-1.5.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mini-swe-agent
-Version: 1.4.2
+Version: 1.5.1
 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
@@ -82,7 +82,7 @@ We now ask: **What if SWE-agent was 100x smaller, and still worked nearly as wel
 
 `mini` is for
 
-- **Researchers** who want to **benchmark, fine-tune or RL** without assumptions, bloat, or surprises
+- **Researchers** who want to **[benchmark](https://swe-bench.com), [fine-tune](https://swesmith.com/) or RL** without assumptions, bloat, or surprises
 - **Developers** who like their tools like their scripts: **short, sharp, and readable**
 - **Engineers** who want something **trivial to sandbox & to deploy anywhere**
 
@@ -90,7 +90,7 @@ Here's some details:
 
 - **Minimal**: Just [100 lines of python](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/agents/default.py) (+100 total for [env](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/environments/local.py),
 [model](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/models/litellm_model.py), [script](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/run/hello_world.py)) — no fancy dependencies!
-- **Powerful:** Resolves 65% of GitHub issues in the [SWE-bench verified benchmark](https://www.swebench.com/) (with Claude Sonnet 4).
+- **Powerful:** Resolves 65% of GitHub issues in the [SWE-bench verified benchmark](https://www.swebench.com/) ([leaderboard](https://swe-bench.com/)).
 - **Convenient:** Comes with UIs that turn this into your daily dev swiss army knife!
 - **Deployable:** In addition to local envs, you can use **docker**, **podman**, **singularity**, **apptainer**, and more
 - **Tested:** [![Codecov](https://img.shields.io/codecov/c/github/swe-agent/mini-swe-agent?style=flat-square)](https://codecov.io/gh/SWE-agent/mini-swe-agent)
@@ -116,6 +116,7 @@ In fact, mini-SWE-agent
 
 This makes it perfect as a baseline system and for a system that puts the language model (rather than
 the agent scaffold) in the middle of our attention.
+You can see the result on the [SWE-bench (bash only)](https://www.swebench.com/) leaderboard, that evaluates the performance of different LMs with `mini`.
 
 </details>
 

{mini_swe_agent-1.4.2 → mini_swe_agent-1.5.1}/README.md

@@ -14,7 +14,7 @@ We now ask: **What if SWE-agent was 100x smaller, and still worked nearly as wel
 
 `mini` is for
 
-- **Researchers** who want to **benchmark, fine-tune or RL** without assumptions, bloat, or surprises
+- **Researchers** who want to **[benchmark](https://swe-bench.com), [fine-tune](https://swesmith.com/) or RL** without assumptions, bloat, or surprises
 - **Developers** who like their tools like their scripts: **short, sharp, and readable**
 - **Engineers** who want something **trivial to sandbox & to deploy anywhere**
 
@@ -22,7 +22,7 @@ Here's some details:
 
 - **Minimal**: Just [100 lines of python](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/agents/default.py) (+100 total for [env](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/environments/local.py),
 [model](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/models/litellm_model.py), [script](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/run/hello_world.py)) — no fancy dependencies!
-- **Powerful:** Resolves 65% of GitHub issues in the [SWE-bench verified benchmark](https://www.swebench.com/) (with Claude Sonnet 4).
+- **Powerful:** Resolves 65% of GitHub issues in the [SWE-bench verified benchmark](https://www.swebench.com/) ([leaderboard](https://swe-bench.com/)).
 - **Convenient:** Comes with UIs that turn this into your daily dev swiss army knife!
 - **Deployable:** In addition to local envs, you can use **docker**, **podman**, **singularity**, **apptainer**, and more
 - **Tested:** [![Codecov](https://img.shields.io/codecov/c/github/swe-agent/mini-swe-agent?style=flat-square)](https://codecov.io/gh/SWE-agent/mini-swe-agent)
@@ -48,6 +48,7 @@ In fact, mini-SWE-agent
 
 This makes it perfect as a baseline system and for a system that puts the language model (rather than
 the agent scaffold) in the middle of our attention.
+You can see the result on the [SWE-bench (bash only)](https://www.swebench.com/) leaderboard, that evaluates the performance of different LMs with `mini`.
 
 </details>
 

{mini_swe_agent-1.4.2 → mini_swe_agent-1.5.1/src/mini_swe_agent.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mini-swe-agent
-Version: 1.4.2
+Version: 1.5.1
 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
@@ -82,7 +82,7 @@ We now ask: **What if SWE-agent was 100x smaller, and still worked nearly as wel
 
 `mini` is for
 
-- **Researchers** who want to **benchmark, fine-tune or RL** without assumptions, bloat, or surprises
+- **Researchers** who want to **[benchmark](https://swe-bench.com), [fine-tune](https://swesmith.com/) or RL** without assumptions, bloat, or surprises
 - **Developers** who like their tools like their scripts: **short, sharp, and readable**
 - **Engineers** who want something **trivial to sandbox & to deploy anywhere**
 
@@ -90,7 +90,7 @@ Here's some details:
 
 - **Minimal**: Just [100 lines of python](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/agents/default.py) (+100 total for [env](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/environments/local.py),
 [model](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/models/litellm_model.py), [script](https://github.com/SWE-agent/mini-swe-agent/blob/main/src/minisweagent/run/hello_world.py)) — no fancy dependencies!
-- **Powerful:** Resolves 65% of GitHub issues in the [SWE-bench verified benchmark](https://www.swebench.com/) (with Claude Sonnet 4).
+- **Powerful:** Resolves 65% of GitHub issues in the [SWE-bench verified benchmark](https://www.swebench.com/) ([leaderboard](https://swe-bench.com/)).
 - **Convenient:** Comes with UIs that turn this into your daily dev swiss army knife!
 - **Deployable:** In addition to local envs, you can use **docker**, **podman**, **singularity**, **apptainer**, and more
 - **Tested:** [![Codecov](https://img.shields.io/codecov/c/github/swe-agent/mini-swe-agent?style=flat-square)](https://codecov.io/gh/SWE-agent/mini-swe-agent)
@@ -116,6 +116,7 @@ In fact, mini-SWE-agent
 
 This makes it perfect as a baseline system and for a system that puts the language model (rather than
 the agent scaffold) in the middle of our attention.
+You can see the result on the [SWE-bench (bash only)](https://www.swebench.com/) leaderboard, that evaluates the performance of different LMs with `mini`.
 
 </details>
 

{mini_swe_agent-1.4.2 → mini_swe_agent-1.5.1}/src/minisweagent/__init__.py

@@ -8,7 +8,7 @@ This file provides:
   unless you want the static type checking.
 """
 
-__version__ = "1.4.2"
+__version__ = "1.5.1"
 
 import os
 from pathlib import Path

{mini_swe_agent-1.4.2 → mini_swe_agent-1.5.1}/src/minisweagent/agents/default.py

@@ -66,8 +66,8 @@ class DefaultAgent:
         cs = asdict(self.config) | asdict(self.env.config) | asdict(self.model.config) | platform.uname()._asdict()
         return Template(template).render(**kwargs, **cs, **os.environ)
 
-    def add_message(self, role: str, content: str):
-        self.messages.append({"role": role, "content": content})
+    def add_message(self, role: str, content: str, **kwargs):
+        self.messages.append({"role": role, "content": content, **kwargs})
 
     def run(self, task: str) -> tuple[str, str]:
         """Run step() until agent is finished. Return exit status & message"""
@@ -92,7 +92,7 @@ class DefaultAgent:
         if 0 < self.config.step_limit <= self.model.n_calls or 0 < self.config.cost_limit <= self.model.cost:
             raise LimitsExceeded()
         response = self.model.query(self.messages)
-        self.add_message("assistant", response["content"])
+        self.add_message("assistant", **response)
         return response
 
     def get_observation(self, response: dict) -> dict:

{mini_swe_agent-1.4.2 → mini_swe_agent-1.5.1}/src/minisweagent/agents/interactive_textual.py

@@ -9,17 +9,19 @@ import re
 import threading
 import time
 import traceback
+from collections.abc import Iterable
 from dataclasses import dataclass, field
 from pathlib import Path
 from typing import Literal
 
 from rich.spinner import Spinner
 from rich.text import Text
-from textual.app import App, ComposeResult
+from textual.app import App, ComposeResult, SystemCommand
 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.screen import Screen
 from textual.widgets import Footer, Header, Input, Static, TextArea
 
 from minisweagent.agents.default import AgentConfig, DefaultAgent, NonTerminatingException, Submitted
@@ -62,11 +64,12 @@ class TextualAgent(DefaultAgent):
             exit_status, result = super().run(task)
         except Exception as e:
             result = str(e)
+            self.app.call_from_thread(self.app.action_quit)
             print(traceback.format_exc())
-            self.app.call_from_thread(self.app.on_agent_finished, "ERROR", result)
             return "ERROR", result
         else:
             self.app.call_from_thread(self.app.on_agent_finished, exit_status, result)
+        self.app.call_from_thread(self.app.action_quit)
         return exit_status, result
 
     def execute_action(self, action: dict) -> dict:
@@ -238,16 +241,22 @@ class SmartInputContainer(Container):
 
 class AgentApp(App):
     BINDINGS = [
-        Binding("right,l", "next_step", "Step++"),
-        Binding("left,h", "previous_step", "Step--"),
-        Binding("0", "first_step", "Step=0"),
-        Binding("$", "last_step", "Step=-1"),
-        Binding("j,down", "scroll_down", "Scroll down"),
-        Binding("k,up", "scroll_up", "Scroll up"),
-        Binding("q", "quit", "Quit"),
-        Binding("y", "yolo", "Switch to YOLO Mode"),
-        Binding("c", "confirm", "Switch to Confirm Mode"),
-        Binding("u", "human", "Switch to Human Mode"),
+        Binding("right,l", "next_step", "Step++", tooltip="Show next step of the agent"),
+        Binding("left,h", "previous_step", "Step--", tooltip="Show previous step of the agent"),
+        Binding("0", "first_step", "Step=0", tooltip="Show first step of the agent", show=False),
+        Binding("$", "last_step", "Step=-1", tooltip="Show last step of the agent", show=False),
+        Binding("j,down", "scroll_down", "Scroll down", show=False),
+        Binding("k,up", "scroll_up", "Scroll up", show=False),
+        Binding("q,ctrl+q", "quit", "Quit", tooltip="Quit the agent"),
+        Binding("y,ctrl+y", "yolo", "YOLO mode", tooltip="Switch to YOLO Mode (LM actions will execute immediately)"),
+        Binding(
+            "c",
+            "confirm",
+            "CONFIRM mode",
+            tooltip="Switch to Confirm Mode (LM proposes commands and you confirm/reject them)",
+        ),
+        Binding("u,ctrl+u", "human", "HUMAN mode", tooltip="Switch to Human Mode (you can now type commands directly)"),
+        Binding("f1,question_mark", "toggle_help_panel", "Help", tooltip="Show help"),
     ]
 
     def __init__(self, model, env, task: str, **kwargs):
@@ -367,6 +376,15 @@ class AgentApp(App):
         except NoMatches:  # might be called when shutting down
             pass
 
+    # --- Other textual overrides ---
+
+    def get_system_commands(self, screen: Screen) -> Iterable[SystemCommand]:
+        # Add to palette
+        yield from super().get_system_commands(screen)
+        for binding in self.BINDINGS:
+            description = f"{binding.description} (shortcut {' OR '.join(binding.key.split(','))})"  # type: ignore[attr-defined]
+            yield SystemCommand(description, binding.tooltip, binding.action)  # type: ignore[attr-defined]
+
     # --- Textual bindings ---
 
     def action_yolo(self):
@@ -404,3 +422,9 @@ class AgentApp(App):
 
     def action_scroll_up(self) -> None:
         self._vscroll.scroll_to(y=self._vscroll.scroll_target_y - 15)
+
+    def action_toggle_help_panel(self) -> None:
+        if self.query("HelpPanel"):
+            self.action_hide_help_panel()
+        else:
+            self.action_show_help_panel()

{mini_swe_agent-1.4.2 → mini_swe_agent-1.5.1}/src/minisweagent/run/extra/config.py

@@ -12,7 +12,7 @@ from dotenv import set_key, unset_key
 from prompt_toolkit import prompt
 from rich.console import Console
 from rich.rule import Rule
-from typer import Option, Typer
+from typer import Argument, Typer
 
 from minisweagent import global_config_file
 
@@ -25,9 +25,7 @@ app = Typer(
 console = Console(highlight=False)
 
 
-_SETUP_HELP = """Welcome to Mini!
-
-To get started, we need to set up your global config file.
+_SETUP_HELP = """To get started, we need to set up your global config file.
 
 You can edit it manually or use the [bold green]mini-extra config set[/bold green] or [bold green]mini-extra config edit[/bold green] commands.
 
@@ -39,6 +37,8 @@ Here's a few popular models and the required API keys:
 [bold green]o3[/bold green] ([bold green]OPENAI_API_KEY[/bold green])
 
 [bold yellow]You can leave any setting blank to skip it.[/bold yellow]
+
+More information at https://mini-swe-agent.com/latest/quickstart/
 """
 
 
@@ -79,16 +79,22 @@ def setup():
 
 @app.command()
 def set(
-    key: str = Option(..., help="The key to set", prompt=True),
-    value: str = Option(..., help="The value to set", prompt=True),
+    key: str | None = Argument(None, help="The key to set"),
+    value: str | None = Argument(None, help="The value to set"),
 ):
     """Set a key in the global config file."""
+    if key is None:
+        key = prompt("Enter the key to set: ")
+    if value is None:
+        value = prompt(f"Enter the value for {key}: ")
     set_key(global_config_file, key, value)
 
 
 @app.command()
-def unset(key: str = Option(..., help="The key to unset")):
+def unset(key: str | None = Argument(None, help="The key to unset")):
     """Unset a key in the global config file."""
+    if key is None:
+        key = prompt("Enter the key to unset: ")
     unset_key(global_config_file, key)
 
 

{mini_swe_agent-1.4.2 → mini_swe_agent-1.5.1}/src/minisweagent/run/mini.py

@@ -73,7 +73,12 @@ def run_textual(model: Model, env: Environment, agent_config: dict, task: str, o
 
 @app.command(help=_HELP_TEXT)
 def main(
-    visual: bool = typer.Option(False, "-v", "--visual", help="Use visual (pager-style) UI (Textual)"),
+    visual: bool = typer.Option(
+        False,
+        "-v",
+        "--visual",
+        help="Toggle (pager-style) UI (Textual) depending on the MSWEA_VISUAL_MODE_DEFAULT environment setting",
+    ),
     model_name: str | None = typer.Option(
         None,
         "-m",
@@ -113,7 +118,8 @@ def main(
     model = get_model(model_name, config.get("model", {}))
     env = LocalEnvironment(**config.get("env", {}))
 
-    if visual:
+    # Both visual flag and the MSWEA_VISUAL_MODE_DEFAULT flip the mode, so it's essentially a XOR
+    if visual == (os.getenv("MSWEA_VISUAL_MODE_DEFAULT", "false") == "false"):
         return run_textual(model, env, config["agent"], task, output)  # type: ignore[arg-type]
     else:
         return run_interactive(model, env, config["agent"], task, output)  # type: ignore[arg-type]

{mini_swe_agent-1.4.2 → mini_swe_agent-1.5.1}/src/minisweagent/run/utils/save.py

@@ -1,18 +1,31 @@
 import json
 from pathlib import Path
 
-from minisweagent import Agent
+from minisweagent import Agent, __version__
 
 
 def save_traj(
     agent: Agent | None,
     path: Path,
     *,
+    print_path: bool = True,
     exit_status: str | None = None,
     result: str | None = None,
     extra_info: dict | None = None,
     **kwargs,
 ):
+    """Save the trajectory of the agent to a file.
+
+    Args:
+        agent: The agent to save the trajectory of.
+        path: The path to save the trajectory to.
+        print_path: Whether to print confirmation of path to the terminal.
+        exit_status: The exit status of the agent.
+        result: The result/submission of the agent.
+        extra_info: Extra information to save (will be merged into the info dict).
+        **kwargs: Additional information to save (will be merged into top level)
+
+    """
     data = {
         "info": {
             "exit_status": exit_status,
@@ -21,6 +34,7 @@ def save_traj(
                 "instance_cost": 0.0,
                 "api_calls": 0,
             },
+            "mini_version": __version__,
         },
         "messages": [],
         "trajectory_format": "mini-swe-agent-1",
@@ -34,3 +48,5 @@ def save_traj(
 
     path.parent.mkdir(parents=True, exist_ok=True)
     path.write_text(json.dumps(data, indent=2))
+    if print_path:
+        print(f"Saved trajectory to '{path}'")