yogsoth-ai/de-anthropocentric-research-engine

collect-independent-judgments

Collect judgments/rankings from multiple genuinely independent evaluators or perspectives before exposing any peer outputs, preserving ballots and rationale for later social-choice aggregation.

Ver código-fonte

Como usar

  1. Copie o comando e verifique o pacote antes de instalar no seu ambiente de agente.
  2. Consulte no original as ferramentas, credenciais e ambientes necessários.
  3. Descreva os dados de entrada e a entrega esperada; confira a saída com as fontes.
npx skills add https://github.com/yogsoth-ai/de-anthropocentric-research-engine --skill collect-independent-judgments

Antes de começar

  • A instalação não inclui assinatura de modelo, franquia API ou acesso aos seus arquivos.
  • Estrelas e instalações indicam atividade, não um teste independente de qualidade ou segurança.
Documento original do Skill

Renderizado do repositório de origem, preservando títulos, exemplos, código, tabelas, links e imagens.

collect-independent-judgments

Purpose

Collect blinded judgments from genuinely independent evaluators or perspectives before exposing peer outputs.

Input contract

yaml
required: [judgment_prompt, evaluator_roster, candidate_set, independence_rules]
optional: [rubric, response_deadline, abstention_policy]
constraints: [evaluators receive no peer judgments; each ballot preserves rationale and provenance]

Procedure

  1. Freeze the candidate set, rubric, prompt, and independence rules before collection.
  2. Obtain one sealed judgment and rationale per evaluator, recording abstentions and unavailable evidence.
  3. Check for shared data, model, prompt, or deliberation leakage before unblinding.
  4. Unblind only after collection, then return ballots, independence flags, and an aggregation-ready dataset.

If independent judgments are complete and their provenance is preserved, consider aggregate-ranking as the next tactic.

Output contract

yaml
produces: [sealed_ballots, independence_audit, rationale_set, aggregation_dataset]
delta_fields: [findings, evidence_updates, uncertainties, decisions, open_questions]

Quality gates

  • A-class gate: declared universe = all commissioned evaluator ballots; numerator = ballots complete, provenance-linked, and leakage-free; batch increment = one sealed ballot; stopping reason = roster exhausted or a declared quorum reached with all exclusions documented; source references = evaluator, prompt, data, and ballot IDs; direction/threshold reason = independence confidence decreases with shared inputs or peer exposure.
  • No aggregation before the leakage check is complete.

Failure and counterexamples

Do not count multiple roles using the same evidence as independent judgments. An abstention is a recorded outcome, not a missing pass.

Provenance map

  • resolved: collective-adjudication

Relacionados por tarefa

do mesmo repositório

Mais Skills

Todos os Skills