release: 0.3.2 with unscripted routing evaluations by phense · Pull Request #140 · phense/engineering-method

phense · 2026-11-13T19:53:36Z

Summary

Syncs Method Core 64d9a02 (72 files): scripts/run_natural_evals.py, and run_host_evals.py no longer lets the host inherit an open stdin (Codex runs timed out without output).
Three unscripted cases under evals/natural/.
Version 0.3.2: manifests, documents, docs/releases/0.3.2.md, backlog EM-022.

No skill changes.

359 tests pass; scripts/validate-plugin exits 0; sync-core --check passes.
Unscripted requests, three runs per case with Writing Method loaded beside this plugin: failing test 2 of 3 on Claude and 3 of 3 on Codex; cross-module feature 3 of 3 on both; typing error (no skill expected) 3 of 3 on Claude and 2 of 3 on Codex. A tendency, not a rate.

🤖 Generated with Claude Code

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>

phense and others added 2 commits November 13, 2026 20:12

test(evals): unscripted routing cases

9223077

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>

release: prepare 0.3.2 with unscripted routing evaluations

8fd705c

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>

phense merged commit ec77b65 into main Nov 13, 2026
2 checks passed

phense deleted the feat/natural-evals branch November 13, 2026 19:55