From b01966bb109291ceecfeceaf3cf7c2d61dac029e Mon Sep 17 00:00:00 2001 From: jared Date: Mon, 6 Jul 2026 17:44:09 -0400 Subject: [PATCH] os-orchestration: trigger-conditioned wording for audit clusters 1-3 Applies the Eval B when->then lesson to the three verified WS1 clusters: explicit model: before every Agent call (cluster 2, cost exposure now that CLAUDE_CODE_SUBAGENT_MODEL is removed), verify resolved model after launch (cluster 1 policy gap), don't re-cover own ground before delegating (cluster 3). Delegation thresholds unchanged. Co-Authored-By: Claude Fable 5 --- plugins/os-orchestration/ORCHESTRATION.md | 18 +++++++++++++----- 1 file changed, 13 insertions(+), 5 deletions(-) diff --git a/plugins/os-orchestration/ORCHESTRATION.md b/plugins/os-orchestration/ORCHESTRATION.md index bbb57f5..8908d2d 100644 --- a/plugins/os-orchestration/ORCHESTRATION.md +++ b/plugins/os-orchestration/ORCHESTRATION.md @@ -3,8 +3,16 @@ - Do single-file, ≤2-tool-call ops directly. Don't delegate them. Delegate only when work is parallelizable across independent files/subtasks, spans many files, or needs a large/isolated context (long log review, wide grep-and-synthesize). -- Every `Agent` spawn passes `model` explicitly. Default `haiku` for mechanical - file-edit/shell work; `sonnet` for anything requiring judgment; `opus` only for - genuinely hard reasoning. -- A short orienting Read before delegating is fine when the target file/path is - uncertain. Don't delegate the orienting step itself. +- Before every `Agent` call → set `model:` explicitly in that call. An omitted + `model` silently bills the subagent at the main-loop model. Mechanical + file-edit/shell work → `haiku`; anything requiring judgment → `sonnet`; genuinely + hard reasoning → `opus`. +- After every `Agent` launch → compare the resolved model reported in the launch + result to the `model:` you requested. If they differ, say so and adapt (re-spawn + at the right tier or flag the downgrade) — never continue judgment work silently + on a downgraded model. +- Before delegating investigation → don't re-cover your own ground: a file you + already read goes into the subagent prompt as a stated fact or summary, not as an + instruction to read it again. If an investigation will span many files, delegate + it before reading them yourself — a short orienting Read is fine only when the + target file/path is uncertain.