Agent
Default implementation lane running GPT-5.6 Luna via the OpenAI Codex CLI (codex exec, reasoning effort max). Route routine, well-specified work here — the spec fully determines the outcome and Codex does the typing at a fraction of the architect's token cost, from a different model family than the session. Receives…
Agent
Second-opinion advisor and final reviewer running Claude's most capable model (Fable 5). Consult at commitment boundaries — before architectural decisions, data migrations, big refactors, or API designs, and whenever the same problem has resisted two attempts — and ALWAYS once at the end of a deliverable, to review…
Agent
High-complexity implementation lane running Claude's most capable model (Fable 5). Route a task here only when the outcome depends heavily on judgment the spec cannot fully capture — subtle concurrency, non-trivial algorithms, security-sensitive paths, gnarly debugging, wide-blast-radius refactors — or when the same…