--- name: arena description: "Spawn N parallel candidates at the same task, pick a base, graft the strongest parts of the losers into it. Use for /arena, 'arena this', 'throw it in the arena', or when one attempt at a non-trivial artifact would lock in the wrong shape." disable-model-invocation: true --- # Arena Fan out N parallel attempts at the same task. Read every candidate end to end. Pick the strongest as the base. Graft the best ideas from the others into it. Verify the synthesized result. ## Start Write a visible checklist with one entry per phase before launching anything. Every launch below follows `../poteto-mode/references/fan-out.md`. 1. Frame 2. Fan out 3. Cross-judge 4. Pick 5. Graft 6. Verify ## Phase A: Frame The N candidates will receive the same prompt, so the prompt is the contract. 1. State the artifact each candidate is producing. 2. Derive the rubric. State what success looks like for *this* task, then turn it into 3-6 concrete gradeable criteria. The rubric is the picker's tool in Phase D. Candidates only see the task. 3. Pick the runners. Runners are diversity seats: each gets an exact authenticated `provider/model-id` from the live catalog, one model family per seat, per `../poteto-mode/references/fan-out.md` section 3, which also covers having fewer families than seats. If a launch rejects a model, run that seat on another model of the same family and say so. Spawn more when the arena covers multiple design directions. Same-family seats, including the same model N times when the work is generation-bound rather than judgment-sensitive, are allowed only when disclosed as context-isolated, not cross-family, and the cross-family result is then reported incomplete. 4. Assign output paths. Each candidate writes to its own location (its own managed `worktree` branch from committed HEAD, per `../poteto-mode/references/fan-out.md` section 5; if the host cannot create one, report that launch as blocked rather than substituting a directory), per the **principle-separate-before-serializing-shared-state** principle skill. ## Phase B: Fan out Launch all N candidates as N independent `subagent` calls in one turn, then end the turn. The host delivers each result. Give each the task, the path to the shared grounding, its own output path, and instructions to produce both the artifact and a short rationale. ```json subagent { "name": "-candidate-1", "task": ". Shared grounding: . Write the artifact and a short rationale to ; the rationale names the alternatives you considered and what you rejected. Commit in your worktree; do not push, merge or open a PR. You are a leaf: launch nothing.", "systemPrompt": "", "model": "/", "thinking": "high", "fork": false, "worktree": { "branch": "-candidate-1" } } ``` Each rationale names the alternatives the candidate considered and what it rejected. If a candidate fails to produce output, proceed with N-1 and note the dropout in the synthesis record. ## Phase C: Cross-judge After all Phase B candidates complete, choose one exact authenticated `provider/model-id` from the live catalog for the cross-judge, a diversity seat. Its family must differ from every candidate author's family, not only the parent's, per `../poteto-mode/references/fan-out.md` section 3; if no such family is authenticated, disclose a same-family context-isolated judge and report the cross-judgment incomplete. Spawn one readonly judge subagent on that model. It sees the rubric and the candidates by path label, scores each criterion, and recommends a base with rationale. It runs in parallel with the parent's reading in Phase D, not with the candidates themselves. Don't spawn the judge while candidates are still writing. ```json subagent { "name": "-cross-judge", "task": "Score candidates against this rubric, criterion by criterion: . Recommend a base with rationale. Read-only. You are a leaf: launch nothing.", "systemPrompt": "", "model": "/", "thinking": "high", "tools": "read, bash", "fork": false } ``` ## Phase D: Pick a base Read every candidate end to end before picking. Score each candidate against the rubric criterion by criterion, not on holistic feel. Compare against the cross-judge. Agreement on the base confirms the pick. Disagreement means one of you is biased or the rubric was ambiguous. Read both rationales before deciding. Pick the base on which candidate a future maintainer can extend most easily without breaking invariants. Prefer the cleaner boundary or smaller API when two feel tied, per the Laziness Protocol. Record the pick and the reason in a short synthesis note alongside the base artifact, including the cross-judge's verdict. ## Phase E: Graft Walk each losing candidate once more and identify what is worth porting into the base. The signal is usually one or two things per candidate, not most of it. Fold each graft in by hand, per the **principle-redesign-from-first-principles** principle skill. Don't paste mechanically. The result has to remain coherent under one mental model. Record what was grafted, from which candidate, and what was rejected and why. When N candidates converge on the same shape, that is a strong agreement signal. Note the convergence in the record and ship the consensus shape. No graft is needed. When N candidates wildly diverge, Phase A was under-specified. Reframe and re-run rather than averaging the divergence. ## Phase F: Verify The synthesized artifact has to hold up under the same scrutiny as any other output, per the **principle-prove-it-works** principle skill. If verification surfaces a problem the arena did not catch, either Phase A was wrong (re-frame and re-run) or one candidate caught it and you missed the graft (go back to Phase E). Don't paper over. ## Outputs One synthesized artifact. One short synthesis note alongside, naming the base, the grafts (with source candidate), the rejections, the dropouts if any, and the verification result.