chore: release 0.72.2

Author: semantic-release

CHANGELOG.md

@@ -1,6 +1,30 @@
 # CHANGELOG
 
 
+## v0.72.2 (2026-03-28)
+
+### Bug Fixes
+
+- Constrained decoding preserves chain-of-thought reasoning
+  ([#200](https://github.com/OpenAdaptAI/openadapt-evals/pull/200),
+  [`d011b2f`](https://github.com/OpenAdaptAI/openadapt-evals/commit/d011b2fcab1a0489ef1e4d3b7d7bc362a6cd64d5))
+
+The Thought/Action format from SYSTEM_PROMPT is now enforced by the constrained decoding regex:
+
+Thought: <up to 500 chars of reasoning>
+
+Action: CLICK(x=0.50, y=0.30)
+
+This gives the model a reasoning budget while guaranteeing parseable output. Prior regex had no
+  prefix (model couldn't reason) or used (.|\n)* (Outlines couldn't compile the DFA).
+
+Also exposes _ACTION_RE (action-only regex) for use by the parser.
+
+Tests updated: 30 pass (was 21).
+
+Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
+
+
 ## v0.72.1 (2026-03-28)
 
 ### Bug Fixes

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "openadapt-evals"
-version = "0.72.1"
+version = "0.72.2"
 description = "Evaluation infrastructure for GUI agent benchmarks"
 readme = "README.md"
 requires-python = ">=3.10"