feat(blitz): spawn-time model assessment over label table

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

skills/blitz/AOE-WORKERS.md

@@ -11,24 +11,32 @@ babysitting and cleanup.
 
 ## Model routing
 
-Default table, by the repo's `difficulty/` label (issues without one: judge
+**Assess each issue at spawn time.** You have just read its body to write the
-the tier yourself from the issue's shape):
+prompt — use that read to pick the model. The question is not "how big is
+this" but **how much judgment does the session still have to exercise**:
 
-| Difficulty | Models |
+- Body settles the approach (root cause named, fix shape decided, numbers
-|---|---|
+  suggested, files pointed at) — the thinking happened at filing time; the
-| hard | Fable (`--tool claude --extra-args "--model claude-fable-5"`) |
+  session executes. **Sonnet** (`--tool claude --extra-args "--model
-| moderate | Opus (`--tool claude --extra-args "--model claude-opus-5"`) |
+  claude-sonnet-5"`), regardless of size: a large mechanical CRUD issue is
-| easy / trivial | Sonnet (`--tool claude --extra-args "--model claude-sonnet-5"`) |
+  still execution.
+- Body states the goal but the session must design the interface, choose the
+  data model, or amend the design doc — **Opus** (`--model claude-opus-5`).
+- The design doc itself is thin or contradictory where this issue lives,
+  correctness is subtle, or the change is cross-cutting with unclear blast
+  radius — **Fable** (`--model claude-fable-5`).
 
-The operator overrides this by just saying so in the invocation ("run these
+A `difficulty/` label is one input — a filing-time guess that cannot see how
-on codex", "use ox alpha for the easy ones") — no config file. Route by how
+much the body scaffolds. Trust your read of the body over it; the label is a
-much judgment the session still has to exercise, not by the label alone: an
+tie-breaker. When in doubt between two tiers take the lower one: escalation
-issue whose body already settles the approach (root cause named, fix shape
+on failure is cheap, and a failed cheap run teaches something a successful
-decided, numbers suggested) drops a tier, because the thinking happened when
+expensive run does not.
-it was filed. When a tier
+
-has two models, alternate within it rather than draining one first — a bad
+The operator overrides any of this by just saying so in the invocation ("run
-run should be visible early. Escalate *sideways* (the tier's other model)
+these on codex", "use ox alpha for the easy ones") — no config file. When
-before escalating up, and never de-escalate mid-issue.
+routing across peer models, alternate rather than draining one first — a bad
+run should be visible early. Escalate *sideways* (a peer model) before
+escalating up, and never de-escalate mid-issue.
 
 ## Spawning