sibyl-supervisor-decision
The sibyl-supervisor-decision skill analyzes experimental results and perspective documents from three debate roles to determine whether a research project should pivot or proceed. It evaluates the experiment summary, optimist, skeptic, and strategist viewpoints, then synthesizes these into a supervisory decision documented in the experiment analysis file. Use this skill when research teams need structured decision-making after experimental phases to guide next steps.
git clone --depth 1 https://github.com/Sibyl-Research-Team/AutoResearch-SibylSystem /tmp/sibyl-supervisor-decision && cp -r /tmp/sibyl-supervisor-decision/.claude/skills/sibyl-supervisor-decision ~/.claude/skills/sibyl-supervisor-decisionSKILL.md
!`SIBYL_WORKSPACE="$ARGUMENTS[0]" .venv/bin/python3 -c "from sibyl.orchestrate import render_skill_prompt; import os; ws = os.environ.get('SIBYL_WORKSPACE', ''); print(render_skill_prompt('supervisor_decision', workspace_path=ws))"`
AGENT_NAME: sibyl-supervisor-decision
AGENT_TIER: sibyl-heavy
SIBYL_ROOT: /Users/cwan0785/sibyl-system
Workspace path: $ARGUMENTS[0]
SPECIAL TASK: Analyze experiment results and the debate opinions.
Read:
- $ARGUMENTS[0]/exp/results/summary.md
- $ARGUMENTS[0]/idea/result_debate/optimist.md
- $ARGUMENTS[0]/idea/result_debate/skeptic.md
- $ARGUMENTS[0]/idea/result_debate/strategist.md
- $ARGUMENTS[0]/idea/proposal.md
Determine: PIVOT or PROCEED?
Write to $ARGUMENTS[0]/supervisor/experiment_analysis.md
End with exactly: DECISION: PIVOT or DECISION: PROCEEDSibyl Codex 独立第三方审查 - 使用 OpenAI Codex 提供不同 AI 视角的评审
Sibyl Codex 写作 agent - 使用 Codex (GPT-5) 撰写论文
Sibyl 比较分析者 agent - 对标 SOTA 和同类工作,定位结果贡献
Sibyl 反对者 agent - 挑战主流假设,寻找反面证据和盲点
Sibyl 批评者 agent - 苛刻但公正的学术批评
Sibyl 编辑 agent - 整合论文各章节为完整稿件
Sibyl 实验主义者 agent - 关注可复现性、数据质量和实验设计严谨性
Sibyl 后台实验监督 agent - 持续刷新 GPU 空闲状态、动态派发排队实验、处理运行时间/状态漂移