v0.2.0

dprovenancekit 0.2.0 — the free, Apache-2.0 reasoning-observability SDK for AI systems. Zero third-party dependencies; Python 3.9+.

Highlights since 0.1.0

• Relicensed BSL 1.1 → Apache-2.0 — the SDK is now permissively licensed for adoption.
• Regression gate in CI — a server-less dprovenancekit gate CLI, plus a drop-in GitHub Action and GitLab CI template that fail a PR/MR when an agent's reasoning drifts from a golden baseline and comment the diff.
• Anomaly rules — ToolDropRule + LoopingRule with a JSON rule registry and a dprovenancekit anomalies subcommand, runnable locally or on every PR.
• Baseline selection — a runs subcommand and two-database gate for picking golden vs candidate.
• Query DSL — new CountStep node.
• Sharper regression detection — reordered critical steps are flagged as HIGH; findings are matched on semantic identity.
• Report export — render_report_html produces a shareable HTML regression report.

Packaging

This package ships only the Apache-2.0 SDK (dprovenancekit). The hosted backend / visualizer is now maintained as a separate commercial product and is not part of this release.

Install

pip install dprovenancekit
pip install "dprovenancekit[langchain]"        # + LangChain adapter
pip install "dprovenancekit[openai-agents]"    # + OpenAI Agents adapter