v2.0.0
Warning
This is the first full release of mini-swe-agent v2. This includes lots of breaking changes.
This is the largest new release since the launch, focusing on supporting more diverse models and switching to toolcalls by default (as opposed to manual parsing of actions).
What's new?
- Tool calls: Native tool calling API is now the support, text-based action parsing is still available
- Multimodal input: Support for images and other content types
- More flexible model support in general
What do I need to change?
I only use the mini CLI with default configs
No changes needed.
I use custom configs
You might need to move some config keys. See Config changes.
I parse/analyze trajectories
Some metadata fields moved to extra. See Trajectory format.
I use the python bindings or built custom subclasses
You will need to refactor. An agent can refactor your code based on the instructions below.
Full migration guide
Read the full migration guide.
Changes compared to v2.0.0a3:
- Enable parallel tool calls and use plain text observation template by @aorwall in #722
- Enh: Only ask for task if unset by @klieret in #738
Changelog compared to last v1
Full Changelog: v1.17.3...v2.0.0