---
title: "mini-swe-agent integration"
description: "Run mini-swe-agent's SWE-bench instances on Promigence sandboxes. Nothing in mini-swe-agent changes: you name the Promigence environment class on the…"
url: "https://www.promigence.ai/integrations/mini-swe-agent"
site: "Promigence"
---

# mini-swe-agent integration

Run mini-swe-agent's SWE-bench instances on Promigence sandboxes. Nothing in mini-swe-agent changes: you name the Promigence environment class on the command line, and each instance runs its commands in its own sandbox while many workers run at once.

The package is `promigence-mini-swe-agent` on PyPI, for Python.

## Install

Install it and set your key.

```bash
pip install promigence-mini-swe-agent  # Python >= 3.10 (mini-swe-agent's floor)
export PROMIGENCE_API_KEY=prom_api_...  PROMIGENCE_API_URL=https://api.promigence.ai/v1
```

## Example

From the package's README on PyPI.

```bash
mini-extra swebench --subset verified --split test --workers 32 \
  -m anthropic/claude-sonnet-4-5 \
  --environment-class promigence_mini_swe_agent.environment.PromigenceEnvironment \
  -c swebench.yaml -c environment.image=docker.io/swebench/sweb.eval.x86_64.django_1776_django-11133:latest \
  -c environment.spend_cap=5
```

## Next

- [promigence-mini-swe-agent on PyPI](https://pypi.org/project/promigence-mini-swe-agent/)
- [All integrations](https://www.promigence.ai/integrations)