From ff420ea3ff56ef393a069a7baa4cb67e693d13c3 Mon Sep 17 00:00:00 2001 From: Jeff Casimir Date: Fri, 28 Aug 2026 17:34:47 -0600 Subject: [PATCH] fix(ce-user-scenarios): sonnet floor for user persona agents User personas exercise features and judge UX -- judgment work, not mechanical extraction. Same policy as the ce-review model-tiering fix on main: orchestrator agent-model wins when defined, sonnet otherwise. This conventional-commit also makes release-please pick up the ce-review/ce-run fixes already on main, which landed without a conventional prefix and were invisible to the release pipeline. Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_014ZyVXdjCUAwHbunnFoREUJ --- plugins/compound-engineering/skills/ce-user-scenarios/SKILL.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/plugins/compound-engineering/skills/ce-user-scenarios/SKILL.md b/plugins/compound-engineering/skills/ce-user-scenarios/SKILL.md index cce600c0b..5bf2cec76 100644 --- a/plugins/compound-engineering/skills/ce-user-scenarios/SKILL.md +++ b/plugins/compound-engineering/skills/ce-user-scenarios/SKILL.md @@ -84,7 +84,7 @@ For each persona file discovered in Step 2: - `{persona_file}` -- the full persona markdown content - `{stage_framing}` -- the stage-specific framing block - `{feature_context}` -- the assembled feature context from Step 3 -4. Spawn a sub-agent with `model: haiku` using the constructed prompt +4. Spawn a sub-agent using the constructed prompt, with `model:` set to the calling orchestrator's `agent-model:` when one is defined, otherwise `sonnet` -- persona evaluation is judgment work, and a persona that misses friction a user would hit costs more than the model tier saves Spawn all persona agents in parallel. If parallel dispatch is not supported, spawn sequentially.