Retire the NextSteps handoff notes and pin the model tiers
The three NextSteps*.md files were per-session handoff notes that had gone stale; the plan they described is superseded. CLAUDE.md gains a MODEL TIERS section recording who is allowed to run as what: Opus coordinates, Sonnet implements, Haiku verifies, and no subagent runs on Opus.
This commit is contained in:
@@ -169,6 +169,21 @@ to paste the **raw output verbatim**, including failures, rather than summarisin
|
||||
reports "all green" without the output is useless — several agents have reported passing gates
|
||||
that were in fact failing.
|
||||
|
||||
### MODEL TIERS (who is allowed to be what)
|
||||
|
||||
Binding for all work in this repo, not just verification.
|
||||
|
||||
| Role | Model | Effort | Allowed to |
|
||||
|---|---|---|---|
|
||||
| Coordinator | **Opus** | **medium** | Dispatch tasks, review returned diffs, make product calls, maintain the plan. **Writes no implementation code.** |
|
||||
| Implementer | **Sonnet** | — | Feature and bugfix work. The only tier allowed to write non-trivial code. |
|
||||
| Verifier / mechanical | **Haiku** | — | Run the full suites and paste raw output; small mechanical edits (renames, i18n string moves, test-fixture churn). |
|
||||
|
||||
* **No subagent may run on Opus.** Not for "a hard task", not for review, not for browser
|
||||
verification. If a task looks too hard for Sonnet, it is too big — split it, do not escalate it.
|
||||
* The coordinator never runs the verification suites itself (see the cost rule above).
|
||||
* Browser verification is delegated too — Sonnet drives Chrome, never the coordinator.
|
||||
|
||||
---
|
||||
|
||||
## ANTI-PATTERNS (CRITICAL GUARDRAILS)
|
||||
|
||||
Reference in New Issue
Block a user