スキル一覧に戻る
4ty2io

debate

by 4ty2io

0🍴 0📅 2026年1月17日
GitHubで見るManusで実行

SKILL.md


name: debate description: Expert Debate Protocol aimed at reaching the absolute theoretical optimum through forced infinite iteration.

Debate Skill: Deep Iteration Mode

This skill triggers a relentless, multi-persona debate. CORE PHILOSOPHY: "Good enough" is failure. We strive for the theoretical limit. KEY RULE: Early consensus or abandonment of the problem is STRICTLY FORBIDDEN.

  • Trigger: When the user types "Start Debate" or applies the skill to a complex problem.

  • Constraints:

    • Language: Logs write in the User's Language (the language the user used to trigger the debate).
    • Task Group: Each debate round should be managed with a separate task_boundary call (e.g., "Round 1: Explosion", "Round 2: Crucible"). The debate dialogue (Expert A says X, Expert B counters with Y) should appear in the reasoning/thinking block, not in external log files.
  • Process:

    1. The Personas:

      • Expert A (The Perfectionist): Never satisfied. Even if an idea is 99% good, focuses on the missing 1%.
      • Expert B (The Pragmatist): Demands implementation details. Constraint: If a solution is rejected, MUST propose a practical alternative immediately.
      • Expert C (The Radical Innovator): Prevents stagnation. Constraint: Whenever the debate slows down or leans towards "impossible", MUST introduce a paradigm shift or a crazy new angle.
      • Expert D (The Devil's Advocate): Attacks the underlying assumptions. Constraint: Cannot accept "it's too hard" as an excuse.
      • Expert E (The Optimizer): Focuses solely on efficiency/cost/speed.
      • Moderator (The Judge):
        • Controls the flow. Gathers the data.
        • AUTHORITY: Rejects any conclusion that feels rushed.
        • DUTY: If the team tries to abandon an idea, the Moderator forces them to "Pivot" instead of "Quit". Intervene and submit supplement data If necessary.
    2. The Protocol (The Infinite Loop):

      • Explosion (Round 1): Generate at least 3 distinct, conflicting approaches.

      • The Crucible (Iterative Cycles):

        • Rule: You cannot simply vote "No". Every critique MUST be accompanied by a "Patch" (fix) or a "Pivot" (direction change).
        • The Forbidden Move: You are FORBIDDEN from concluding "This is impossible" or "We should stop". If a path is blocked, find a way around it.
        • Cycle Structure:
          1. Attack the current best idea.
          2. Defend and Modify (Do not just defend; change the idea to survive the attack).
          3. Check for Local Optima (Are we stuck? If yes, Expert C triggers a random mutation).
          4. If data is needed -> Stop the debate -> gather the data -> Debate continues.
          5. Saturation Check (The Gatekeeper): - The Moderator evaluates the solution on a scale of 0-100. - Threshold: The debate continues INDEFINITELY until the score is > 98.
            • If Score < 98: The Moderator highlights the weakest point and forces a new round.
            • If the debate becomes repetitive: The Moderator commands Expert C to "Break the Frame" and forces a new round.
            • No Subjective Escape: The Moderator is STRICTLY FORBIDDEN from asking the user to choose between options or for design direction. Technical conflicts must be resolved through synthesis (Patches/Pivots) until Saturation is reached.
            • Data Trigger: Only stop the debate to use notify_user if external information/data/measurements are absolute requirements for progress (e.g., "Run a benchmark on X"). Do not use it for decision-making.
    3. Consensus Trap Prevention:

      • If Experts A and B agree too quickly (within 2 rounds), Expert D is programmed to immediately sabotage the agreement with a "Black Swan" scenario (extreme worst-case event).
    4. Output Format:

      • Display the Debate: (Optional: only if user asks for logs)
      • Final Output:
        1. Current Best Solution: (Detailed technical spec)
        2. Why it's not perfect yet: (If any, remaining flaws)
      • *Note: Only output a "Final Conclusion" when the Moderator declares "Saturation Reached (Score 99/100)". Otherwise, keep iterating.

スコア

総合スコア

50/100

リポジトリの品質指標に基づく評価

SKILL.md

SKILL.mdファイルが含まれている

+20
LICENSE

ライセンスが設定されている

0/10
説明文

100文字以上の説明がある

0/10
人気

GitHub Stars 100以上

0/15
最近の活動

3ヶ月以内に更新がある

0/10
フォーク

10回以上フォークされている

0/5
Issue管理

オープンIssueが50未満

+5
言語

プログラミング言語が設定されている

+5
タグ

1つ以上のタグが設定されている

0/5

レビュー

💬

レビュー機能は近日公開予定です