v0.2.0

What's Changed

• docs: README with generated side-by-side demo GIF, badges, PyPI quickstart by @abrichr in #5
• docs: tighten README by @abrichr in #6
• feat: compiled-replay vs computer-use-agent benchmark by @abrichr in #7
• docs: add compiled-vs-agent benchmark to README by @abrichr in #8
• feat: OpenEMR public-demo spike — scroll, resolution hardening, showcase by @abrichr in #9
• feat: closed-loop scrolling — SCROLL steps scroll until the next anchor resolves; OpenEMR 5/5 by @abrichr in #10
• feat: OpenEMR flagship benchmark with hard cost guardrails by @abrichr in #11
• docs: adversarial validation suite — failure-mode matrix and LIMITS by @abrichr in #12
• fix: identity-verified click targets and verified typed input — close the silent wrong-action modes by @abrichr in #13
• fix: stability-selected postcondition mining and parameter hygiene by @abrichr in #15
• feat: hybrid compiled+agent-fallback benchmark by @abrichr in #14
• feat: DOM-selector benchmark arm — the incumbent comparison by @abrichr in #17
• fix: identity matcher rebuilt against a frozen adversarial corpus (3rd wrong-patient reopening) + armed-coverage metric by @abrichr in #16
• feat: WindowsBackend + desktop recording-adapter contract (desktop spike phase 1) by @abrichr in #19
• feat: automated Parallels desktop benchmark pipeline (Phase 2) by @abrichr in #20
• chore: release v0.2.0 — safety fixes, link LIMITS above the fold by @abrichr in #22

Full Changelog: v0.1.0...v0.2.0

v0.1.0

What's Changed

• feat: record→compile→replay→heal reference implementation (v0) by @abrichr in #1
• feat(cli): replay self-serves MockMed when --url is omitted; add --drift by @abrichr in #2
• chore: genericize layered-platform references ahead of public release by @abrichr in #3
• feat: PyPI release workflow (trusted publishing); playwright as core dependency by @abrichr in #4

New Contributors

• @abrichr made their first contribution in #1

Full Changelog: https://github.com/OpenAdaptAI/openadapt-flow/commits/v0.1.0