Project Pivot Planner

Decide what to do when the original research direction is not working. This skill is for mid-project decisions — after enough evidence exists to rule out engineering bugs and still the core claim is not holding.

Use this skill when:

• multiple experiments consistently fail to support the core claim despite fixing engineering bugs
• an adversarial reviewer successfully argues that the main contribution is not novel
• the primary baseline unexpectedly closes the gap to the proposed method
• a related work publication makes the original contribution redundant
• advisor/committee feedback consistently questions the project's direction
• the project has been in result-diagnosis for more than 2–3 cycles without resolution

Do not use this skill for per-experiment debugging — use experiment-debugger. Do not use this skill when the issue is a specific surprising result — use result-diagnosis. Do not use this skill for new project validation — use research-idea-validator.

Pair this skill with: