Tool profile
SWE-agent
Princeton NLP
An open-source research agent for software engineering tasks such as issue resolution and repository-level problem solving.
What it's used for
SWE-agent is used to experiment with agentic software engineering, benchmark task completion, and automate controlled coding workflows over repositories.
Categories
How you access it
See whether you access this through a vendor-hosted app, managed cloud, or official client.
APIHigh complexityUsage based
Connected model APIs
Teams often pair SWE-agent with hosted model providers for reasoning and code generation.
How you deploy or integrate it
See whether you can self-host it, deploy it in your stack, or integrate it through APIs and runtimes.
LocalHigh complexitySelf-hosted infra cost
Research workflow on local infrastructure
The framework is typically run on user-controlled machines, VMs, or clusters for experiments and evaluation.