pivot-or-kill

Escalate research decisions after three consecutive core experiment failures.

11|1|Updated Feb 22, 2026
One-click install
npx skills add https://github.com/EvoClaw/amplify --skill pivot-or-kill
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: pivot-or-kill
Source: https://github.com/EvoClaw/amplify/tree/main/skills/pivot-or-kill
Command: npx skills add https://github.com/EvoClaw/amplify --skill pivot-or-kill

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Repeated failures in core experiments stall progress; Pivot-or-Kill enforces honest status checks and user escalation to decide whether to pivot, downgrade, or kill.

Core Features & Use Cases

  • Honest failure summarization after three consecutive unsuccessful experiments.
  • Root-cause analysis classification and triage into one of three options.
  • User-driven decision flow with explicit pivot/downgrade/kill choices and gates.

Quick Start

Trigger the flow after three consecutive core experiment failures to summarize status, analyze root causes, and present three escalation options for user decision.

Frequently Asked Questions about pivot-or-kill

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate project termination decisions after repeated experiment failures?

Automate project termination decisions by enforcing a structured escalation flow after three consecutive experiment failures, triggering an honest status summary, root-cause analysis, and three user-driven options to pivot, downgrade, or kill.

When should I escalate a pivot decision in iterative AI research workflows?

Escalate a pivot decision in iterative AI research workflows exactly when three consecutive core experiment failures occur, shifting the workflow from autonomous execution to a structured human-in-the-loop decision gate.

What is the best way to structure root-cause analysis for failed AI experiments?

The best way to structure root-cause analysis for failed AI experiments is to classify failures and triage them into three explicit options: pivoting the method, downgrading targets, or terminating the project for user selection.

How do I implement decision gates to stop agents from running failed experiments indefinitely?

Implement decision gates by setting a hard threshold of three consecutive core experiment failures, which immediately halts autonomous execution and enforces a three-step analysis before any further research actions proceed.

Can I use automated escalation flows for non-AI iterative research projects?

Automated escalation flows can be applied to any iterative research workflow where agents must choose between pivoting methods, downgrading targets, or terminating projects after repeated consecutive failures.

What are the limitations of using a three-strike failure threshold for research pivots?

The limitation of a three-strike failure threshold is that it is a rigid trigger; it does not evaluate the severity or partial success of individual experiments, strictly enforcing escalation only after three consecutive core failures.