v0.2.26 3.4.0

Name: hello-algo
Author: krahets

high risk

krahets/hello-algoSep 25, 2026 → Sep 25, 20260 days between releases

What's New

Pentest Swarm AI adds exploit chains; DSPy introduces Jev integration and native LM engines

Breaking Changes

Experimental LM types from DSPy 3.3 are replaced
RLM's interpreter_factory input name is now reserved

New Features

Embedded library of 13 exploit chains in Pentest Swarm AI
Chain compiler CLI (list/info/run) in Pentest Swarm AI
Native LM engines via dspy.lm15 in DSPy
Jev typed decision types (Noul, Choice, Score) in DSPy
ReAnchor optimizer for calibrating decisions in DSPy

Improvements

On-demand registry for chain updates in Pentest Swarm AI
Async ReActV2 support in DSPy

Bug Fixes

Updated docs and reconciled StyleSmuggler in Pentest Swarm AI

Upgrade Notes

For DSPy, migrate to new LM types by version 3.5 and use interpreter_factory as a keyword argument. Pentest Swarm AI introduces a new CLI for managing chains.

The Two Releases

v0.2.26

Sep 25, 2026· 27 days ago

Introduces exploit-chain schema, an embedded library of 13 famous chains, a chain compiler CLI, and an on-demand registry.

Full release notes

3.4.0

Sep 25, 2026· 27 days ago

Introduces Jev integration via TypeSafe with typed decision types (Noul, Choice, Score), native LM engines bundled with DSPy, and the ReAnchor optimizer for calibrating decisions against metrics.

Full release notes