mini-swe-agent 1.14.4__tar.gz → 1.16.0__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {mini_swe_agent-1.14.4/src/mini_swe_agent.egg-info → mini_swe_agent-1.16.0}/PKG-INFO +18 -15
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/README.md +16 -14
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/pyproject.toml +1 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0/src/mini_swe_agent.egg-info}/PKG-INFO +18 -15
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/mini_swe_agent.egg-info/SOURCES.txt +5 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/mini_swe_agent.egg-info/requires.txt +1 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/__init__.py +1 -1
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/agents/default.py +3 -3
- mini_swe_agent-1.16.0/src/minisweagent/config/extra/swebench_xml.yaml +215 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/environments/docker.py +7 -1
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/models/__init__.py +3 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/models/extra/roulette.py +2 -3
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/models/litellm_model.py +18 -8
- mini_swe_agent-1.16.0/src/minisweagent/models/litellm_response_api_model.py +80 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/models/openrouter_model.py +9 -8
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/models/portkey_model.py +38 -30
- mini_swe_agent-1.16.0/src/minisweagent/models/portkey_response_api_model.py +74 -0
- mini_swe_agent-1.16.0/src/minisweagent/models/requesty_model.py +119 -0
- mini_swe_agent-1.16.0/src/minisweagent/models/utils/openai_utils.py +41 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/run/extra/config.py +2 -2
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/run/mini.py +1 -2
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/run/utils/save.py +3 -1
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/LICENSE.md +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/setup.cfg +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/mini_swe_agent.egg-info/dependency_links.txt +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/mini_swe_agent.egg-info/entry_points.txt +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/mini_swe_agent.egg-info/top_level.txt +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/__main__.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/agents/__init__.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/agents/interactive.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/agents/interactive_textual.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/config/README.md +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/config/__init__.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/config/default.yaml +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/config/extra/__init__.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/config/extra/swebench.yaml +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/config/extra/swebench_roulette.yaml +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/config/github_issue.yaml +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/config/mini.tcss +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/config/mini.yaml +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/config/mini_no_temp.yaml +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/environments/__init__.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/environments/extra/__init__.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/environments/extra/bubblewrap.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/environments/extra/swerex_docker.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/environments/local.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/environments/singularity.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/models/anthropic.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/models/extra/__init__.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/models/test_models.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/models/utils/__init__.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/models/utils/cache_control.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/models/utils/key_per_thread.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/py.typed +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/run/__init__.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/run/extra/__init__.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/run/extra/swebench.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/run/extra/swebench_single.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/run/extra/utils/__init__.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/run/extra/utils/batch_progress.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/run/github_issue.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/run/hello_world.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/run/inspector.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/run/mini_extra.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/run/utils/__init__.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/utils/__init__.py +0 -0
- {mini_swe_agent-1.14.4 → mini_swe_agent-1.16.0}/src/minisweagent/utils/log.py +0 -0
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.4
|
|
2
2
|
Name: mini-swe-agent
|
|
3
|
-
Version: 1.
|
|
3
|
+
Version: 1.16.0
|
|
4
4
|
Summary: Nano SWE Agent - A simple AI software engineering agent
|
|
5
5
|
Author-email: Kilian Lieret <kilian.lieret@posteo.de>, "Carlos E. Jimenez" <carlosej@princeton.edu>
|
|
6
6
|
License: MIT License
|
|
@@ -66,6 +66,7 @@ Requires-Dist: mike; extra == "dev"
|
|
|
66
66
|
Requires-Dist: mkdocs-material; extra == "dev"
|
|
67
67
|
Requires-Dist: mkdocs-glightbox; extra == "dev"
|
|
68
68
|
Requires-Dist: mkdocs-redirects; extra == "dev"
|
|
69
|
+
Requires-Dist: portkey-ai; extra == "dev"
|
|
69
70
|
Dynamic: license-file
|
|
70
71
|
|
|
71
72
|
<div align="center">
|
|
@@ -74,6 +75,7 @@ Dynamic: license-file
|
|
|
74
75
|
|
|
75
76
|
# The 100 line AI agent that solves GitHub issues & more
|
|
76
77
|
|
|
78
|
+
📣 [Gemini 3 Pro reaches 74% on SWE-bench verified with mini-swe-agent!](https://x.com/KLieret/status/1991164693839270372)<br/>
|
|
77
79
|
📣 [New blogpost: Randomly switching between GPT-5 and Sonnet 4 boosts performance](https://www.swebench.com/SWE-bench/blog/2025/08/19/mini-roulette/)
|
|
78
80
|
|
|
79
81
|
[](https://mini-swe-agent.com/latest/)
|
|
@@ -94,7 +96,7 @@ Here's some details:
|
|
|
94
96
|
|
|
95
97
|
- **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),
|
|
96
98
|
[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!
|
|
97
|
-
- **Powerful:** Resolves >
|
|
99
|
+
- **Powerful:** Resolves >74% of GitHub issues in the [SWE-bench verified benchmark](https://www.swebench.com/) ([leaderboard](https://swe-bench.com/)).
|
|
98
100
|
- **Convenient:** Comes with UIs that turn this into your daily dev swiss army knife!
|
|
99
101
|
- **Deployable:** In addition to local envs, you can use **docker**, **podman**, **singularity**, **apptainer**, and more
|
|
100
102
|
- **Tested:** [](https://codecov.io/gh/SWE-agent/mini-swe-agent)
|
|
@@ -248,7 +250,7 @@ agent.run("Write a sudoku game")
|
|
|
248
250
|
|
|
249
251
|
## Let's get started!
|
|
250
252
|
|
|
251
|
-
Option 1
|
|
253
|
+
**Option 1:** If you just want to try out the CLI (package installed in anonymous virtual environment)
|
|
252
254
|
|
|
253
255
|
```bash
|
|
254
256
|
pip install uv && uvx mini-swe-agent [-v]
|
|
@@ -256,19 +258,19 @@ pip install uv && uvx mini-swe-agent [-v]
|
|
|
256
258
|
pip install pipx && pipx ensurepath && pipx run mini-swe-agent [-v]
|
|
257
259
|
```
|
|
258
260
|
|
|
259
|
-
Option 2
|
|
261
|
+
**Option 2:** Install CLI & python bindings in current environment
|
|
260
262
|
|
|
261
263
|
```bash
|
|
262
|
-
pip install mini-swe-agent
|
|
264
|
+
pip install mini-swe-agent
|
|
265
|
+
mini -v # run the CLI
|
|
263
266
|
```
|
|
264
267
|
|
|
265
|
-
Option 3
|
|
268
|
+
**Option 3:** Install from source (developer setup)
|
|
266
269
|
|
|
267
270
|
```bash
|
|
268
271
|
git clone https://github.com/SWE-agent/mini-swe-agent.git
|
|
269
|
-
cd mini-swe-agent
|
|
270
|
-
|
|
271
|
-
mini [-v]
|
|
272
|
+
cd mini-swe-agent && pip install -e .
|
|
273
|
+
mini [-v] # run the CLI
|
|
272
274
|
```
|
|
273
275
|
|
|
274
276
|
Read more in our [documentation](https://mini-swe-agent.com/latest/):
|
|
@@ -298,14 +300,15 @@ If you found this work helpful, please consider citing the [SWE-agent paper](htt
|
|
|
298
300
|
Our other projects:
|
|
299
301
|
|
|
300
302
|
<div align="center">
|
|
301
|
-
<a href="https://github.com/SWE-agent/SWE-agent"><img src="https://
|
|
303
|
+
<a href="https://github.com/SWE-agent/SWE-agent"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/sweagent_logo_text_below.svg" alt="SWE-agent" height="120px"></a>
|
|
302
304
|
|
|
303
|
-
<a href="https://github.com/SWE-agent/SWE-ReX"><img src="https://
|
|
305
|
+
<a href="https://github.com/SWE-agent/SWE-ReX"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/swerex_logo_text_below.svg" alt="SWE-ReX" height="120px"></a>
|
|
304
306
|
|
|
305
|
-
<a href="https://github.com/SWE-bench/SWE-bench"><img src="https://
|
|
307
|
+
<a href="https://github.com/SWE-bench/SWE-bench"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/swebench_logo_text_below.svg" alt="SWE-bench" height="120px"></a>
|
|
306
308
|
|
|
307
|
-
<a href="https://github.com/SWE-bench/SWE-smith"><img src="https://
|
|
309
|
+
<a href="https://github.com/SWE-bench/SWE-smith"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/swesmith_logo_text_below.svg" alt="SWE-smith" height="120px"></a>
|
|
308
310
|
|
|
309
|
-
<a href="https://github.com/
|
|
311
|
+
<a href="https://github.com/codeclash-ai/codeclash"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/codeclash_logo_text_below.svg" alt="CodeClash" height="120px"></a>
|
|
312
|
+
|
|
313
|
+
<a href="https://github.com/SWE-bench/sb-cli"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/sbcli_logo_text_below.svg" alt="sb-cli" height="120px"></a>
|
|
310
314
|
</div>
|
|
311
|
-
|
|
@@ -4,6 +4,7 @@
|
|
|
4
4
|
|
|
5
5
|
# The 100 line AI agent that solves GitHub issues & more
|
|
6
6
|
|
|
7
|
+
📣 [Gemini 3 Pro reaches 74% on SWE-bench verified with mini-swe-agent!](https://x.com/KLieret/status/1991164693839270372)<br/>
|
|
7
8
|
📣 [New blogpost: Randomly switching between GPT-5 and Sonnet 4 boosts performance](https://www.swebench.com/SWE-bench/blog/2025/08/19/mini-roulette/)
|
|
8
9
|
|
|
9
10
|
[](https://mini-swe-agent.com/latest/)
|
|
@@ -24,7 +25,7 @@ Here's some details:
|
|
|
24
25
|
|
|
25
26
|
- **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),
|
|
26
27
|
[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!
|
|
27
|
-
- **Powerful:** Resolves >
|
|
28
|
+
- **Powerful:** Resolves >74% of GitHub issues in the [SWE-bench verified benchmark](https://www.swebench.com/) ([leaderboard](https://swe-bench.com/)).
|
|
28
29
|
- **Convenient:** Comes with UIs that turn this into your daily dev swiss army knife!
|
|
29
30
|
- **Deployable:** In addition to local envs, you can use **docker**, **podman**, **singularity**, **apptainer**, and more
|
|
30
31
|
- **Tested:** [](https://codecov.io/gh/SWE-agent/mini-swe-agent)
|
|
@@ -178,7 +179,7 @@ agent.run("Write a sudoku game")
|
|
|
178
179
|
|
|
179
180
|
## Let's get started!
|
|
180
181
|
|
|
181
|
-
Option 1
|
|
182
|
+
**Option 1:** If you just want to try out the CLI (package installed in anonymous virtual environment)
|
|
182
183
|
|
|
183
184
|
```bash
|
|
184
185
|
pip install uv && uvx mini-swe-agent [-v]
|
|
@@ -186,19 +187,19 @@ pip install uv && uvx mini-swe-agent [-v]
|
|
|
186
187
|
pip install pipx && pipx ensurepath && pipx run mini-swe-agent [-v]
|
|
187
188
|
```
|
|
188
189
|
|
|
189
|
-
Option 2
|
|
190
|
+
**Option 2:** Install CLI & python bindings in current environment
|
|
190
191
|
|
|
191
192
|
```bash
|
|
192
|
-
pip install mini-swe-agent
|
|
193
|
+
pip install mini-swe-agent
|
|
194
|
+
mini -v # run the CLI
|
|
193
195
|
```
|
|
194
196
|
|
|
195
|
-
Option 3
|
|
197
|
+
**Option 3:** Install from source (developer setup)
|
|
196
198
|
|
|
197
199
|
```bash
|
|
198
200
|
git clone https://github.com/SWE-agent/mini-swe-agent.git
|
|
199
|
-
cd mini-swe-agent
|
|
200
|
-
|
|
201
|
-
mini [-v]
|
|
201
|
+
cd mini-swe-agent && pip install -e .
|
|
202
|
+
mini [-v] # run the CLI
|
|
202
203
|
```
|
|
203
204
|
|
|
204
205
|
Read more in our [documentation](https://mini-swe-agent.com/latest/):
|
|
@@ -228,14 +229,15 @@ If you found this work helpful, please consider citing the [SWE-agent paper](htt
|
|
|
228
229
|
Our other projects:
|
|
229
230
|
|
|
230
231
|
<div align="center">
|
|
231
|
-
<a href="https://github.com/SWE-agent/SWE-agent"><img src="https://
|
|
232
|
+
<a href="https://github.com/SWE-agent/SWE-agent"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/sweagent_logo_text_below.svg" alt="SWE-agent" height="120px"></a>
|
|
232
233
|
|
|
233
|
-
<a href="https://github.com/SWE-agent/SWE-ReX"><img src="https://
|
|
234
|
+
<a href="https://github.com/SWE-agent/SWE-ReX"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/swerex_logo_text_below.svg" alt="SWE-ReX" height="120px"></a>
|
|
234
235
|
|
|
235
|
-
<a href="https://github.com/SWE-bench/SWE-bench"><img src="https://
|
|
236
|
+
<a href="https://github.com/SWE-bench/SWE-bench"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/swebench_logo_text_below.svg" alt="SWE-bench" height="120px"></a>
|
|
236
237
|
|
|
237
|
-
<a href="https://github.com/SWE-bench/SWE-smith"><img src="https://
|
|
238
|
+
<a href="https://github.com/SWE-bench/SWE-smith"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/swesmith_logo_text_below.svg" alt="SWE-smith" height="120px"></a>
|
|
238
239
|
|
|
239
|
-
<a href="https://github.com/
|
|
240
|
+
<a href="https://github.com/codeclash-ai/codeclash"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/codeclash_logo_text_below.svg" alt="CodeClash" height="120px"></a>
|
|
241
|
+
|
|
242
|
+
<a href="https://github.com/SWE-bench/sb-cli"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/sbcli_logo_text_below.svg" alt="sb-cli" height="120px"></a>
|
|
240
243
|
</div>
|
|
241
|
-
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.4
|
|
2
2
|
Name: mini-swe-agent
|
|
3
|
-
Version: 1.
|
|
3
|
+
Version: 1.16.0
|
|
4
4
|
Summary: Nano SWE Agent - A simple AI software engineering agent
|
|
5
5
|
Author-email: Kilian Lieret <kilian.lieret@posteo.de>, "Carlos E. Jimenez" <carlosej@princeton.edu>
|
|
6
6
|
License: MIT License
|
|
@@ -66,6 +66,7 @@ Requires-Dist: mike; extra == "dev"
|
|
|
66
66
|
Requires-Dist: mkdocs-material; extra == "dev"
|
|
67
67
|
Requires-Dist: mkdocs-glightbox; extra == "dev"
|
|
68
68
|
Requires-Dist: mkdocs-redirects; extra == "dev"
|
|
69
|
+
Requires-Dist: portkey-ai; extra == "dev"
|
|
69
70
|
Dynamic: license-file
|
|
70
71
|
|
|
71
72
|
<div align="center">
|
|
@@ -74,6 +75,7 @@ Dynamic: license-file
|
|
|
74
75
|
|
|
75
76
|
# The 100 line AI agent that solves GitHub issues & more
|
|
76
77
|
|
|
78
|
+
📣 [Gemini 3 Pro reaches 74% on SWE-bench verified with mini-swe-agent!](https://x.com/KLieret/status/1991164693839270372)<br/>
|
|
77
79
|
📣 [New blogpost: Randomly switching between GPT-5 and Sonnet 4 boosts performance](https://www.swebench.com/SWE-bench/blog/2025/08/19/mini-roulette/)
|
|
78
80
|
|
|
79
81
|
[](https://mini-swe-agent.com/latest/)
|
|
@@ -94,7 +96,7 @@ Here's some details:
|
|
|
94
96
|
|
|
95
97
|
- **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),
|
|
96
98
|
[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!
|
|
97
|
-
- **Powerful:** Resolves >
|
|
99
|
+
- **Powerful:** Resolves >74% of GitHub issues in the [SWE-bench verified benchmark](https://www.swebench.com/) ([leaderboard](https://swe-bench.com/)).
|
|
98
100
|
- **Convenient:** Comes with UIs that turn this into your daily dev swiss army knife!
|
|
99
101
|
- **Deployable:** In addition to local envs, you can use **docker**, **podman**, **singularity**, **apptainer**, and more
|
|
100
102
|
- **Tested:** [](https://codecov.io/gh/SWE-agent/mini-swe-agent)
|
|
@@ -248,7 +250,7 @@ agent.run("Write a sudoku game")
|
|
|
248
250
|
|
|
249
251
|
## Let's get started!
|
|
250
252
|
|
|
251
|
-
Option 1
|
|
253
|
+
**Option 1:** If you just want to try out the CLI (package installed in anonymous virtual environment)
|
|
252
254
|
|
|
253
255
|
```bash
|
|
254
256
|
pip install uv && uvx mini-swe-agent [-v]
|
|
@@ -256,19 +258,19 @@ pip install uv && uvx mini-swe-agent [-v]
|
|
|
256
258
|
pip install pipx && pipx ensurepath && pipx run mini-swe-agent [-v]
|
|
257
259
|
```
|
|
258
260
|
|
|
259
|
-
Option 2
|
|
261
|
+
**Option 2:** Install CLI & python bindings in current environment
|
|
260
262
|
|
|
261
263
|
```bash
|
|
262
|
-
pip install mini-swe-agent
|
|
264
|
+
pip install mini-swe-agent
|
|
265
|
+
mini -v # run the CLI
|
|
263
266
|
```
|
|
264
267
|
|
|
265
|
-
Option 3
|
|
268
|
+
**Option 3:** Install from source (developer setup)
|
|
266
269
|
|
|
267
270
|
```bash
|
|
268
271
|
git clone https://github.com/SWE-agent/mini-swe-agent.git
|
|
269
|
-
cd mini-swe-agent
|
|
270
|
-
|
|
271
|
-
mini [-v]
|
|
272
|
+
cd mini-swe-agent && pip install -e .
|
|
273
|
+
mini [-v] # run the CLI
|
|
272
274
|
```
|
|
273
275
|
|
|
274
276
|
Read more in our [documentation](https://mini-swe-agent.com/latest/):
|
|
@@ -298,14 +300,15 @@ If you found this work helpful, please consider citing the [SWE-agent paper](htt
|
|
|
298
300
|
Our other projects:
|
|
299
301
|
|
|
300
302
|
<div align="center">
|
|
301
|
-
<a href="https://github.com/SWE-agent/SWE-agent"><img src="https://
|
|
303
|
+
<a href="https://github.com/SWE-agent/SWE-agent"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/sweagent_logo_text_below.svg" alt="SWE-agent" height="120px"></a>
|
|
302
304
|
|
|
303
|
-
<a href="https://github.com/SWE-agent/SWE-ReX"><img src="https://
|
|
305
|
+
<a href="https://github.com/SWE-agent/SWE-ReX"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/swerex_logo_text_below.svg" alt="SWE-ReX" height="120px"></a>
|
|
304
306
|
|
|
305
|
-
<a href="https://github.com/SWE-bench/SWE-bench"><img src="https://
|
|
307
|
+
<a href="https://github.com/SWE-bench/SWE-bench"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/swebench_logo_text_below.svg" alt="SWE-bench" height="120px"></a>
|
|
306
308
|
|
|
307
|
-
<a href="https://github.com/SWE-bench/SWE-smith"><img src="https://
|
|
309
|
+
<a href="https://github.com/SWE-bench/SWE-smith"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/swesmith_logo_text_below.svg" alt="SWE-smith" height="120px"></a>
|
|
308
310
|
|
|
309
|
-
<a href="https://github.com/
|
|
311
|
+
<a href="https://github.com/codeclash-ai/codeclash"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/codeclash_logo_text_below.svg" alt="CodeClash" height="120px"></a>
|
|
312
|
+
|
|
313
|
+
<a href="https://github.com/SWE-bench/sb-cli"><img src="https://raw.githubusercontent.com/SWE-agent/swe-agent-media/refs/heads/main/media/logos_banners/sbcli_logo_text_below.svg" alt="sb-cli" height="120px"></a>
|
|
310
314
|
</div>
|
|
311
|
-
|
|
@@ -24,6 +24,7 @@ src/minisweagent/config/mini_no_temp.yaml
|
|
|
24
24
|
src/minisweagent/config/extra/__init__.py
|
|
25
25
|
src/minisweagent/config/extra/swebench.yaml
|
|
26
26
|
src/minisweagent/config/extra/swebench_roulette.yaml
|
|
27
|
+
src/minisweagent/config/extra/swebench_xml.yaml
|
|
27
28
|
src/minisweagent/environments/__init__.py
|
|
28
29
|
src/minisweagent/environments/docker.py
|
|
29
30
|
src/minisweagent/environments/local.py
|
|
@@ -34,14 +35,18 @@ src/minisweagent/environments/extra/swerex_docker.py
|
|
|
34
35
|
src/minisweagent/models/__init__.py
|
|
35
36
|
src/minisweagent/models/anthropic.py
|
|
36
37
|
src/minisweagent/models/litellm_model.py
|
|
38
|
+
src/minisweagent/models/litellm_response_api_model.py
|
|
37
39
|
src/minisweagent/models/openrouter_model.py
|
|
38
40
|
src/minisweagent/models/portkey_model.py
|
|
41
|
+
src/minisweagent/models/portkey_response_api_model.py
|
|
42
|
+
src/minisweagent/models/requesty_model.py
|
|
39
43
|
src/minisweagent/models/test_models.py
|
|
40
44
|
src/minisweagent/models/extra/__init__.py
|
|
41
45
|
src/minisweagent/models/extra/roulette.py
|
|
42
46
|
src/minisweagent/models/utils/__init__.py
|
|
43
47
|
src/minisweagent/models/utils/cache_control.py
|
|
44
48
|
src/minisweagent/models/utils/key_per_thread.py
|
|
49
|
+
src/minisweagent/models/utils/openai_utils.py
|
|
45
50
|
src/minisweagent/run/__init__.py
|
|
46
51
|
src/minisweagent/run/github_issue.py
|
|
47
52
|
src/minisweagent/run/hello_world.py
|
|
@@ -2,7 +2,6 @@
|
|
|
2
2
|
|
|
3
3
|
import re
|
|
4
4
|
import subprocess
|
|
5
|
-
from collections.abc import Callable
|
|
6
5
|
from dataclasses import asdict, dataclass
|
|
7
6
|
|
|
8
7
|
from jinja2 import StrictUndefined, Template
|
|
@@ -25,6 +24,7 @@ class AgentConfig:
|
|
|
25
24
|
)
|
|
26
25
|
format_error_template: str = "Please always provide EXACTLY ONE action in triple backticks."
|
|
27
26
|
action_observation_template: str = "Observation: {{output}}"
|
|
27
|
+
action_regex: str = r"```bash\s*\n(.*?)\n```"
|
|
28
28
|
step_limit: int = 0
|
|
29
29
|
cost_limit: float = 3.0
|
|
30
30
|
|
|
@@ -54,7 +54,7 @@ class LimitsExceeded(TerminatingException):
|
|
|
54
54
|
|
|
55
55
|
|
|
56
56
|
class DefaultAgent:
|
|
57
|
-
def __init__(self, model: Model, env: Environment, *, config_class:
|
|
57
|
+
def __init__(self, model: Model, env: Environment, *, config_class: type = AgentConfig, **kwargs):
|
|
58
58
|
self.config = config_class(**kwargs)
|
|
59
59
|
self.messages: list[dict] = []
|
|
60
60
|
self.model = model
|
|
@@ -106,7 +106,7 @@ class DefaultAgent:
|
|
|
106
106
|
|
|
107
107
|
def parse_action(self, response: dict) -> dict:
|
|
108
108
|
"""Parse the action from the message. Returns the action."""
|
|
109
|
-
actions = re.findall(
|
|
109
|
+
actions = re.findall(self.config.action_regex, response["content"], re.DOTALL)
|
|
110
110
|
if len(actions) == 1:
|
|
111
111
|
return {"action": actions[0].strip(), **response}
|
|
112
112
|
raise FormatError(self.render_template(self.config.format_error_template, actions=actions))
|
|
@@ -0,0 +1,215 @@
|
|
|
1
|
+
agent:
|
|
2
|
+
system_template: |
|
|
3
|
+
You are a helpful assistant that can interact multiple times with a computer shell to solve programming tasks.
|
|
4
|
+
Your response must contain exactly ONE bash code block with ONE command (or commands connected with && or ||).
|
|
5
|
+
|
|
6
|
+
Include a THOUGHT section before your command where you explain your reasoning process.
|
|
7
|
+
Format your response as shown in <format_example>.
|
|
8
|
+
|
|
9
|
+
<format_example>
|
|
10
|
+
THOUGHT: Your reasoning and analysis here
|
|
11
|
+
|
|
12
|
+
<bash_code>your_command_here</bash_code>
|
|
13
|
+
</format_example>
|
|
14
|
+
|
|
15
|
+
Failure to follow these rules will cause your response to be rejected.
|
|
16
|
+
instance_template: |
|
|
17
|
+
<pr_description>
|
|
18
|
+
Consider the following PR description:
|
|
19
|
+
{{task}}
|
|
20
|
+
</pr_description>
|
|
21
|
+
|
|
22
|
+
<instructions>
|
|
23
|
+
# Task Instructions
|
|
24
|
+
|
|
25
|
+
## Overview
|
|
26
|
+
You're a software engineer interacting continuously with a computer by submitting commands.
|
|
27
|
+
You'll be helping implement necessary changes to meet requirements in the PR description.
|
|
28
|
+
Your task is specifically to make changes to non-test files in the current directory in order to fix the issue described in the PR description in a way that is general and consistent with the codebase.
|
|
29
|
+
|
|
30
|
+
IMPORTANT: This is an interactive process where you will think and issue ONE command, see its result, then think and issue your next command.
|
|
31
|
+
|
|
32
|
+
For each response:
|
|
33
|
+
1. Include a THOUGHT section explaining your reasoning and what you're trying to accomplish
|
|
34
|
+
2. Provide exactly ONE bash command to execute
|
|
35
|
+
|
|
36
|
+
## Important Boundaries
|
|
37
|
+
- MODIFY: Regular source code files in /testbed (this is the working directory for all your subsequent commands)
|
|
38
|
+
- DO NOT MODIFY: Tests, configuration files (pyproject.toml, setup.cfg, etc.)
|
|
39
|
+
|
|
40
|
+
## Recommended Workflow
|
|
41
|
+
1. Analyze the codebase by finding and reading relevant files
|
|
42
|
+
2. Create a script to reproduce the issue
|
|
43
|
+
3. Edit the source code to resolve the issue
|
|
44
|
+
4. Verify your fix works by running your script again
|
|
45
|
+
5. Test edge cases to ensure your fix is robust
|
|
46
|
+
|
|
47
|
+
## Command Execution Rules
|
|
48
|
+
You are operating in an environment where
|
|
49
|
+
1. You write a single command
|
|
50
|
+
2. The system executes that command in a subshell
|
|
51
|
+
3. You see the result
|
|
52
|
+
4. You write your next command
|
|
53
|
+
|
|
54
|
+
Each response should include:
|
|
55
|
+
1. A **THOUGHT** section where you explain your reasoning and plan
|
|
56
|
+
2. A single bash code block with your command
|
|
57
|
+
|
|
58
|
+
Format your responses like included within the <format_example> block:
|
|
59
|
+
|
|
60
|
+
<format_example>
|
|
61
|
+
THOUGHT: Here I explain my reasoning process, analysis of the current situation,
|
|
62
|
+
and what I'm trying to accomplish with the command below.
|
|
63
|
+
|
|
64
|
+
<bash_code>your_command_here </bash_code></format_example>
|
|
65
|
+
|
|
66
|
+
**CRITICAL REQUIREMENTS:**
|
|
67
|
+
- Your response SHOULD include a THOUGHT section explaining your reasoning
|
|
68
|
+
- Your response MUST include EXACTLY ONE bash code block
|
|
69
|
+
- This bash block MUST contain EXACTLY ONE command (or a set of commands connected with && or ||)
|
|
70
|
+
- If you include zero or multiple bash blocks, or no command at all, YOUR RESPONSE WILL FAIL
|
|
71
|
+
- Do NOT try to run multiple independent commands in separate blocks in one response
|
|
72
|
+
- Directory or environment variable changes are not persistent. Every action is executed in a new subshell.
|
|
73
|
+
- However, you can prefix any action with `MY_ENV_VAR=MY_VALUE cd /path/to/working/dir && ...` or write/load environment variables from files
|
|
74
|
+
|
|
75
|
+
Example of a CORRECT response:
|
|
76
|
+
|
|
77
|
+
<example_response>
|
|
78
|
+
THOUGHT: I need to understand the structure of the repository first. Let me check what files are in the current directory to get a better understanding of the codebase.
|
|
79
|
+
|
|
80
|
+
<bash_code>ls -la</bash_code>
|
|
81
|
+
</example_response>
|
|
82
|
+
|
|
83
|
+
Example of an INCORRECT response:
|
|
84
|
+
|
|
85
|
+
<example_response>
|
|
86
|
+
THOUGHT: I need to examine the codebase and then look at a specific file. I'll run multiple commands to do this.
|
|
87
|
+
|
|
88
|
+
<bash_code>ls -la</bash_code>
|
|
89
|
+
|
|
90
|
+
Now I'll read the file:
|
|
91
|
+
|
|
92
|
+
<bash_code>cat file.txt</bash_code>
|
|
93
|
+
</example_response>
|
|
94
|
+
|
|
95
|
+
If you need to run multiple commands, either:
|
|
96
|
+
|
|
97
|
+
1. Combine them in one block using && or ||
|
|
98
|
+
|
|
99
|
+
<bash_code>command1 && command2 || echo "Error occurred"</bash_code>
|
|
100
|
+
|
|
101
|
+
2. Wait for the first command to complete, see its output, then issue the next command in your following response.
|
|
102
|
+
|
|
103
|
+
## Environment Details
|
|
104
|
+
|
|
105
|
+
- You have a full Linux shell environment
|
|
106
|
+
- Always use non-interactive flags (-y, -f) for commands
|
|
107
|
+
- Avoid interactive tools like vi, nano, or any that require user input
|
|
108
|
+
- If a command isn't available, you can install it
|
|
109
|
+
|
|
110
|
+
## Useful Command Examples
|
|
111
|
+
|
|
112
|
+
### Create a new file:
|
|
113
|
+
|
|
114
|
+
<bash_code>cat <<'EOF' > newfile.py
|
|
115
|
+
import numpy as np
|
|
116
|
+
hello = "world"
|
|
117
|
+
print(hello)
|
|
118
|
+
EOF</bash_code>
|
|
119
|
+
|
|
120
|
+
### Edit files with sed:
|
|
121
|
+
|
|
122
|
+
Replace all occurrences
|
|
123
|
+
|
|
124
|
+
<bash_code>sed -i 's/old_string/new_string/g' filename.py</bash_code>
|
|
125
|
+
|
|
126
|
+
Replace only first occurrence
|
|
127
|
+
|
|
128
|
+
<bash_code>sed -i 's/old_string/new_string/' filename.py</bash_code>
|
|
129
|
+
|
|
130
|
+
Replace first occurrence on line 1
|
|
131
|
+
|
|
132
|
+
<bash_code>sed -i '1s/old_string/new_string/' filename.py</bash_code>
|
|
133
|
+
|
|
134
|
+
Replace all occurrences in lines 1-10
|
|
135
|
+
|
|
136
|
+
<bash_code>sed -i '1,10s/old_string/new_string/g' filename.py</bash_code>
|
|
137
|
+
|
|
138
|
+
### View file content:
|
|
139
|
+
|
|
140
|
+
View specific lines with numbers
|
|
141
|
+
|
|
142
|
+
<bash_code> nl -ba filename.py | sed -n '10,20p'</bash_code>
|
|
143
|
+
|
|
144
|
+
Any other command you want to run
|
|
145
|
+
|
|
146
|
+
<bash_code>anything</bash_code>
|
|
147
|
+
|
|
148
|
+
## Submission
|
|
149
|
+
|
|
150
|
+
When you've completed your work (reading, editing, testing), and cannot make further progress
|
|
151
|
+
issue exactly the following command:
|
|
152
|
+
|
|
153
|
+
<bash_code>echo COMPLETE_TASK_AND_SUBMIT_FINAL_OUTPUT && git add -A && git diff --cached</bash_code>
|
|
154
|
+
|
|
155
|
+
This command will submit your work.
|
|
156
|
+
You cannot continue working (reading, editing, testing) in any way on this task after submitting.
|
|
157
|
+
</instructions>
|
|
158
|
+
action_observation_template: |
|
|
159
|
+
<returncode>{{output.returncode}}</returncode>
|
|
160
|
+
{% if output.output | length < 10000 -%}
|
|
161
|
+
<output>
|
|
162
|
+
{{ output.output -}}
|
|
163
|
+
</output>
|
|
164
|
+
{%- else -%}
|
|
165
|
+
<warning>
|
|
166
|
+
The output of your last command was too long.
|
|
167
|
+
Please try a different command that produces less output.
|
|
168
|
+
If you're looking at a file you can try use head, tail or sed to view a smaller number of lines selectively.
|
|
169
|
+
If you're using grep or find and it produced too much output, you can use a more selective search pattern.
|
|
170
|
+
If you really need to see something from the full command's output, you can redirect output to a file and then search in that file.
|
|
171
|
+
</warning>
|
|
172
|
+
{%- set elided_chars = output.output | length - 10000 -%}
|
|
173
|
+
<output_head>
|
|
174
|
+
{{ output.output[:5000] }}
|
|
175
|
+
</output_head>
|
|
176
|
+
<elided_chars>
|
|
177
|
+
{{ elided_chars }} characters elided
|
|
178
|
+
</elided_chars>
|
|
179
|
+
<output_tail>
|
|
180
|
+
{{ output.output[-5000:] }}
|
|
181
|
+
</output_tail>
|
|
182
|
+
{%- endif -%}
|
|
183
|
+
format_error_template: |
|
|
184
|
+
Please always provide EXACTLY ONE action in the `<bash_code>` block, found {{actions|length}} actions.
|
|
185
|
+
|
|
186
|
+
Please format your action in a `<bash_code>` block as shown in <response_example>.
|
|
187
|
+
|
|
188
|
+
<response_example>
|
|
189
|
+
Here are some thoughts about why you want to perform the action.
|
|
190
|
+
|
|
191
|
+
<bash_code>ls -la</bash_code>
|
|
192
|
+
</response_example>
|
|
193
|
+
|
|
194
|
+
If you have completed your assignment, please consult the first message about how to
|
|
195
|
+
submit your solution (you will not be able to continue working on this task after that).
|
|
196
|
+
step_limit: 250
|
|
197
|
+
cost_limit: 3.
|
|
198
|
+
action_regex: <bash_code>(.*?)</bash_code>
|
|
199
|
+
|
|
200
|
+
environment:
|
|
201
|
+
cwd: "/testbed"
|
|
202
|
+
timeout: 60
|
|
203
|
+
env:
|
|
204
|
+
PAGER: cat
|
|
205
|
+
MANPAGER: cat
|
|
206
|
+
LESS: -R
|
|
207
|
+
PIP_PROGRESS_BAR: 'off'
|
|
208
|
+
TQDM_DISABLE: '1'
|
|
209
|
+
environment_class: docker
|
|
210
|
+
|
|
211
|
+
model:
|
|
212
|
+
model_name: "minimax/minimax-m2"
|
|
213
|
+
model_class: openrouter
|
|
214
|
+
model_kwargs:
|
|
215
|
+
temperature: 0.0
|
|
@@ -34,7 +34,13 @@ class DockerEnvironmentConfig:
|
|
|
34
34
|
|
|
35
35
|
|
|
36
36
|
class DockerEnvironment:
|
|
37
|
-
def __init__(
|
|
37
|
+
def __init__(
|
|
38
|
+
self,
|
|
39
|
+
*,
|
|
40
|
+
config_class: type = DockerEnvironmentConfig,
|
|
41
|
+
logger: logging.Logger | None = None,
|
|
42
|
+
**kwargs,
|
|
43
|
+
):
|
|
38
44
|
"""This class executes bash commands in a Docker container using direct docker commands.
|
|
39
45
|
See `DockerEnvironmentConfig` for keyword arguments.
|
|
40
46
|
"""
|
|
@@ -81,8 +81,11 @@ def get_model_name(input_model_name: str | None = None, config: dict | None = No
|
|
|
81
81
|
_MODEL_CLASS_MAPPING = {
|
|
82
82
|
"anthropic": "minisweagent.models.anthropic.AnthropicModel",
|
|
83
83
|
"litellm": "minisweagent.models.litellm_model.LitellmModel",
|
|
84
|
+
"litellm_response": "minisweagent.models.litellm_response_api_model.LitellmResponseAPIModel",
|
|
84
85
|
"openrouter": "minisweagent.models.openrouter_model.OpenRouterModel",
|
|
85
86
|
"portkey": "minisweagent.models.portkey_model.PortkeyModel",
|
|
87
|
+
"portkey_response": "minisweagent.models.portkey_response_api_model.PortkeyResponseAPIModel",
|
|
88
|
+
"requesty": "minisweagent.models.requesty_model.RequestyModel",
|
|
86
89
|
"deterministic": "minisweagent.models.test_models.DeterministicModel",
|
|
87
90
|
}
|
|
88
91
|
|
|
@@ -1,5 +1,4 @@
|
|
|
1
1
|
import random
|
|
2
|
-
from collections.abc import Callable
|
|
3
2
|
from dataclasses import asdict, dataclass
|
|
4
3
|
|
|
5
4
|
from minisweagent import Model
|
|
@@ -14,7 +13,7 @@ class RouletteModelConfig:
|
|
|
14
13
|
|
|
15
14
|
|
|
16
15
|
class RouletteModel:
|
|
17
|
-
def __init__(self, *, config_class:
|
|
16
|
+
def __init__(self, *, config_class: type = RouletteModelConfig, **kwargs):
|
|
18
17
|
"""This "meta"-model randomly selects one of the models at every call"""
|
|
19
18
|
self.config = config_class(**kwargs)
|
|
20
19
|
self.models = [get_model(config=config) for config in self.config.model_kwargs]
|
|
@@ -50,7 +49,7 @@ class InterleavingModelConfig:
|
|
|
50
49
|
|
|
51
50
|
|
|
52
51
|
class InterleavingModel(RouletteModel):
|
|
53
|
-
def __init__(self, *, config_class:
|
|
52
|
+
def __init__(self, *, config_class: type = InterleavingModelConfig, **kwargs):
|
|
54
53
|
"""This "meta"-model alternates between the models in the sequence for every call"""
|
|
55
54
|
super().__init__(config_class=config_class, **kwargs)
|
|
56
55
|
|