From 6550706ce31e92ce1680a7006325e255777d97ca Mon Sep 17 00:00:00 2001 From: Betty Date: Fri, 14 Aug 2026 18:38:41 +0200 Subject: [PATCH] post: commit a1f1eeb in myai-orchestrator --- .../commits/2026-08-14-commit-a1f1eeb.md | 45 +++++++++++++++++++ 1 file changed, 45 insertions(+) create mode 100644 content/posts/commits/2026-08-14-commit-a1f1eeb.md diff --git a/content/posts/commits/2026-08-14-commit-a1f1eeb.md b/content/posts/commits/2026-08-14-commit-a1f1eeb.md new file mode 100644 index 0000000..00bcf5a --- /dev/null +++ b/content/posts/commits/2026-08-14-commit-a1f1eeb.md @@ -0,0 +1,45 @@ +--- +title: "[myai-orchestrator] fix(orchestrator): simplify clean_code() and increase MAX_NEW_TOKENS to 256" +date: 2026-08-14T18:38:41+02:00 +draft: false +tags: ["commit", "myai-orchestrator", "main"] +categories: ["Git Activity"] +summary: "Commit a1f1eeb par Betty dans myai-orchestrator" +author: "Betty" +--- + +## Commit `a1f1eeb` + +| | | +|---|---| +| **Repository** | myai-orchestrator | +| **Branch** | `main` | +| **Author** | Betty | +| **Hash** | `a1f1eeb136017425bcbd9cf0fb8b00feea4a5aff` | + + +### Description + +clean_code() was too aggressive, stripping valid Python code by tracking +an in_code flag that could fail to trigger. New logic: extract from +markdown fences (regex), then find first import/from line and take +everything from there. Much simpler and more reliable. + +MAX_NEW_TOKENS 128→256 to reduce truncation-induced SyntaxErrors. + +Co-Authored-By: Claude Sonnet 4.6 + +### Files Changed + +``` +M myai_orchestrator +M stack/myai-orchestrator.yml +``` + +### Diff Summary + +``` + myai_orchestrator | 39 +++++++++++++++------------------------ + stack/myai-orchestrator.yml | 2 +- + 2 files changed, 16 insertions(+), 25 deletions(-) +```