failure future reset

The problem

LLM agents often fall into loops of safe half-work, shipping slop, or stalling before consequential decisions. Without a forcing function for accountability, agents default to the appearance of work rather than driving actual results.

What it does

• Evaluates work across multiple time horizons (1 week to 10 years) to predict specific failure outcomes.
• Forces the agent to document concrete negative consequences rather than vague warnings.
• Requires exactly one owned change with a specific owner, deadline, and verification method.
• Standardizes failure reporting for multi-agent handoffs to ensure peers understand the stakes.

Why this beats prompting it yourself

Basic prompting fails to stop agent drift because it lacks a structured ritual for self-correction. This skill provides a specific framework for "Failure Pictures" that anchors the agent's logic in long-term outcomes, preventing the habit of shipping mediocre code or content just to close a ticket.

Use cases

• Pre-deployment checks for high-stakes infrastructure changes.
• Breaking creative blocks when an agent is stuck in repetitive output loops.
• Handoffs between specialized agents to ensure mission alignment.
• Budget-sensitive operations where "safe" paths lead to wasted spend.