From 2020e63bd2af86f0366eb0eb329aed788a0cb913 Mon Sep 17 00:00:00 2001 From: Peter Steinberger Date: Tue, 21 Apr 2026 13:18:10 +0100 Subject: [PATCH] test: harden repo contract qa scenario --- .../agents/instruction-followthrough-repo-contract.md | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) diff --git a/qa/scenarios/agents/instruction-followthrough-repo-contract.md b/qa/scenarios/agents/instruction-followthrough-repo-contract.md index 8a7d756d298..bc52f64f7d1 100644 --- a/qa/scenarios/agents/instruction-followthrough-repo-contract.md +++ b/qa/scenarios/agents/instruction-followthrough-repo-contract.md @@ -49,8 +49,14 @@ execution: Evidence path: AGENT.md -> SOUL.md -> FOLLOWTHROUGH_INPUT.md -> repo-contract-summary.txt prompt: |- Repo contract followthrough check. Read AGENT.md, SOUL.md, and FOLLOWTHROUGH_INPUT.md first. - Then follow the repo contract exactly, write ./repo-contract-summary.txt, and reply with - three labeled lines: Read, Wrote, Status. + Then use the write tool to create ./repo-contract-summary.txt with this exact body: + + Repo contract + Evidence path: AGENT.md -> SOUL.md -> FOLLOWTHROUGH_INPUT.md -> repo-contract-summary.txt + Status: complete + + Do not send the final reply until ./repo-contract-summary.txt exists. After writing it, reply with + three labeled lines only: Read, Wrote, Status. Do not stop after planning and do not ask for permission before the first feasible action. expectedReplyAll: - "read:"