wbj0256/dsh-agent-handoff-deliberation ↗★ 0
@wbj0256/dsh-agent-handoff-deliberation
Privacy-first handoffs and structured multi-agent deliberation for DeepSeek Harness integrations.
AI 分析
提供 HandoffLedger 和 DeliberationRoom 模块，用于管理多智能体之间的任务交接与结构化辩论。适合需要构建复杂多智能体协同、决策研讨系统的开发者。
インストール
検証済み bundle がないか、互換性チェックに失敗しています。先にリポジトリの説明を読んでください。 README 全文を読む ↗
ドキュメント
README 全文を読む ↗Usage
import { HandoffLedger, DeliberationRoom } from '@wbj0256/dsh-agent-handoff-deliberation'
const handoffs = new HandoffLedger()
handoffs.create({ id: 'design-7', from: 'planner', to: 'builder', task: 'Implement the validated API contract.' })
handoffs.accept('design-7', 'builder')
handoffs.complete('design-7', 'builder', { artifact: 'commit reference' })
const debate = new DeliberationRoom()
debate.open('release-1', 'Release now?', ['builder', 'reviewer', 'lead'])
debate.position('release-1', 'builder', 'Release after tests.', ['unit tests passed'])
debate.position('release-1', 'reviewer', 'Hold for the abuse-case test.', ['threat model'])
debate.rebut('release-1', 'builder', 'reviewer', 'The boundary test covers this case.')
debate.decide('release-1', 'lead', 'Release only with boundary evidence recorded.')