Flowgentic 0.3.1

Flowgentic 0.3.1 is a patch release. It fixes the defects that the 0.3.0 documentation listed, in the memory strategies, MutableGraph, the execution report and telemetry. Flowgentic stays in maintenance mode; see the project status.

Fixed

• Memory, trim_middle: with one free place beside the system messages, every message was kept. The strategy now fills every free place.
• Memory, importance_based: after the first trim, each message was scored by another message's importance. The scores now follow the trimmed history, which also corrects get_relevant_context.
• Memory, summarize: the summary could take the history one message over the limit. It now fits, and a failed summary trims as trim_last does.
• MutableGraph.update_graph returns False for a replacement node that is already active, instead of listing it twice and failing the rebuild.
• generate_execution_artifacts(generate_report=...) was inverted. True now writes the report, and remains the default.
• Telemetry: tool calls are read from AIMessage.tool_calls (and invalid_tool_calls), so models other than OpenAI's appear in the report; an AI message's text is recorded as the node's final response even without additional_kwargs.

The changelog has the details.

Tested environment

Unchanged from 0.3.0: LangGraph 1.2.16, langchain-core 1.6.13, langchain 1.4.7, radical-asyncflow 0.6.0 and academy-py 0.3.0, from uv.lock, on Python 3.10, 3.11 and 3.12. Local backends only.

Install

The wheel and sdist are attached below; Flowgentic is not on PyPI.

uv venv --python 3.10 flowgentic-env
uv pip install --python flowgentic-env/bin/python https://github.com/stride-research/flowgentic/releases/download/v0.3.1/flowgentic-0.3.1-py3-none-any.whl

Known limitations

• DEBUG logging is on by default, and config.yml is looked up in the working directory at import.
• ADR mode is not supported: --controller adr needs an ADR source checkout.
• Not validated: an engine that the caller creates, and any remote or HPC backend. Cancellation of in-flight AsyncFlow tasks is validated on local backends only.
• The adapter's default retry (3 attempts of at most 30 s each) also wraps graph nodes: a node that runs longer than 30 s fails, and a node that fails with a retryable error re-runs its tasks.
• The agentic demo records no token usage, even in live mode.
• The design-pattern examples need a model provider and credentials, and CI does not run them.
• The Academy integration (flowgentic.academy) has no tests, and its examples fail at import.

Links: documentation · substitution experiment · 0.3.0 release