devlisted

> **Model routing for this tool** (from `model-routing.data.yaml`, snapshot 2026-08-17): > `frontier` → **fable** · `workhorse` → **opus** · `standard` → **sonnet** · `utility` → **haiku** > Roles → tiers: ba-challenger: standard · ba-draft: standard · dev-r1: workhorse · dev-r2: standard (high-stakes: workhorse) · dev-r3: workhorse · explore: utility · qa-challenger: standard · sa-background: workhorse · sa-challenger: workhorse · sa-writer: workhorse > Resolve at runtime: `python3 .vteam/scripts/model_route.py <role> --tool claude-code [--high-stakes]` — high-stakes diffs (review.high_stakes_*) bump dev-r2 to the workhorse tier. > Spawning a subagent: pass the resolved name as the Agent tool's `model` parameter. # /dev — implement a ticket, prove it before delivering **Immutable principles** (violation = the task is not done): 1. **The spec is the spec, the ticket is a claim.** Expected behavior comes from the spec shard (`docs/specs/<feature>.md`) and the schema — never from ticket prose alone. Ticket contradicts spec/schema → STOP and ask before coding. 2. **No code before understanding.** T1's task-sheet (requirement, acceptance criteria, touched models/routes, cited sources) must exist before the first edit. Ambiguous ticket → ask, do not guess. 3. **Minimal, surgical change** (the `guidelines` workflow is the method authority — load it). No drive-by refactors, no scope expansion; a side finding gets its own note/ticket (T5), never smuggled into thi