From 27ecfb707d2c30e4c80e7d05e4e051c0a18c681c Mon Sep 17 00:00:00 2001 From: Nick Shirokov Date: Thu, 7 May 2026 21:27:08 +0300 Subject: [PATCH] =?UTF-8?q?revert(plugins):=20drop=20UTF-8=20BOM=20on=20SK?= =?UTF-8?q?ILL.md=20=E2=80=94=20Codex=20YAML=20parser=20breaks=20on=20it?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit BOM before `---` makes Codex's strict frontmatter parser reject every SKILL.md ("Skipped loading 66 skill(s)"). Without BOM the skills load and execute correctly; the only remaining issue is mojibake display of Cyrillic in SKILL.md previews — that's a Codex rendering bug, not ours. Co-Authored-By: Claude Opus 4.7 (1M context) --- .github/workflows/build-ports.yml | 9 --------- 1 file changed, 9 deletions(-) diff --git a/.github/workflows/build-ports.yml b/.github/workflows/build-ports.yml index c0015228..db55eb73 100644 --- a/.github/workflows/build-ports.yml +++ b/.github/workflows/build-ports.yml @@ -206,15 +206,6 @@ jobs: sed -e "s|{{PLUGIN_NAME}}|${PLUGIN_NAME}|g" \ .github/templates/claude-plugin.json.tmpl > build/.claude-plugin/plugin.json - - name: Add UTF-8 BOM to SKILL.md (Codex reads Cyrillic as CP1252 without BOM) - if: matrix.platform == 'codex' - run: | - find build -name 'SKILL.md' -print0 | while IFS= read -r -d '' f; do - if ! head -c 3 "$f" | od -An -tx1 | tr -d ' \n' | grep -q '^efbbbf'; then - { printf '\xef\xbb\xbf'; cat "$f"; } > "$f.tmp" && mv "$f.tmp" "$f" - fi - done - - name: Copy LICENSE run: cp LICENSE build/LICENSE