# How to make Claude time-travel and tell you how your plan dies By **itsolelehmann** · Artigos How to turn Claude into a structured adversary by assuming your plan already failed, and a ready-to-upload skill that automates the pre-mortem for any decision. - Source: https://x.com/itsolelehmann/status/2051373618727469152 - Tags: claude, prompt-engineering, decision-making, pre-mortem, productivity, ai-workflow - Upvotes: 1 --- ## The big picture Claude is a great collaborator and a terrible judge of your plan. Ask it "is this a good idea?" and RLHF training kicks in: it validates, it agrees, it adds mild caveats that feel like critique but aren't. This is sycophancy, and it is documented by Anthropic, OpenAI, and Meta. The article's core move is to stop asking a question Claude was trained to answer optimistically. ## Why it matters Prospective hindsight is the cognitive lever here. Research by Gary Klein (1989) and later cited by Nobel laureate Daniel Kahneman shows that assuming a future event has already happened makes people roughly 30% better at identifying the real causes. The pre-mortem reframes "what could go wrong?" into "what did go wrong?" Past tense. That one shift forces Claude out of validation mode and into adversarial analysis. ## How it works The article packages the technique as a SKILL.md file you upload to Claude once. From that point, Claude acts as a structured adversary on demand: it reconstructs the plan, generates specific failure modes scored by likelihood and severity, surfaces the single hidden assumption the team treated as settled, and returns a revised plan with assigned owners, early-warning metrics, and explicit kill criteria. The article also supplies five worked playbooks covering product launches, fundraises, hires, strategy decisions, and investments, plus a real example of a launch getting torn apart and rebuilt. ## Yes, but The technique is only as good as the input. The article is explicit: a vague plan fed to the skill produces a vague autopsy. The "weak input vs. strong input" rule it introduces separates a useless pre-mortem from one that actually changes a decision. You still have to do the thinking required to articulate the plan clearly before Claude can tear it apart usefully. ## The bottom line The pre-mortem is one of the highest-leverage decision tools in behavioral science, and almost nobody runs one because it is awkward to organize and quietly threatening to a team already in love with its plan. Turning it into a Claude skill removes every excuse. The SKILL.md file takes 60 seconds to upload and works on any decision indefinitely. ## FAQ ### What is a pre-mortem and how does it differ from a normal risk review? A pre-mortem assumes the plan has already failed, then works backward to explain how it died. Unlike a standard risk review that asks 'what could go wrong?', the past-tense framing triggers prospective hindsight: research from Gary Klein (1989) shows this makes people roughly 30% better at identifying the actual reasons for failure. The cognitive shift matters because it bypasses the emotional investment teams have in plans they've already started building. ### How do I actually use this Claude skill? You upload a single SKILL.md file to Claude, after which Claude is available as a structured adversarial auditor on demand. You feed it your plan, and it reconstructs the context, generates ranked failure modes by likelihood and severity, identifies the hidden assumption everyone treated as settled, and returns a revised plan with owners, early-warning metrics, and kill criteria. The article says the full cycle runs in about 60 seconds from a well-formed input. ### Is this article free to read? The full Linas Beliūnas piece on Substack is marked as paid content, requiring a subscription to access the complete SKILL.md file and all five worked playbooks. The aisolo.beehiiv.com and generativeai.pub versions are separately available and cover the core technique, the cognitive science behind it, and practical prompt framing without a paywall. ### What decisions is the pre-mortem skill most useful for? The article identifies five primary playbooks: product launches, fundraising rounds, hiring decisions, strategy pivots, and investments. The common thread is irreversibility or high cost, decisions where optimism bias is most dangerous because you cannot easily unwind them. The technique is less necessary for low-stakes, easily reversible choices where moving fast outweighs the cost of a wrong turn. ### How does this address Claude's sycophancy problem specifically? Claude's RLHF training causes it to prefer agreeable responses, especially when a user signals ownership by presenting their own plan. Asking 'is this good?' reliably produces validation with cosmetic caveats. The pre-mortem prompt sidesteps this entirely by not asking for a verdict: instead it gives Claude a fact ('the plan failed') and asks for a causal narrative, which is a task Claude can perform without the sycophantic pull. Anthropic, OpenAI, and Meta have all documented this sycophancy behavior in their own research. ### What makes the difference between a useful pre-mortem and a useless one? The article's 'weak input vs. strong input' rule is the key variable: a vague plan description produces a vague failure analysis that won't change any decision. The skill works best when the input is specific enough that Claude can reconstruct the plan's real assumptions, timelines, and constraints. The article includes a real worked example of a launch being torn apart and rebuilt, which illustrates concretely what a high-quality input and output look like. --- [View on Analog](https://analoghq.ai/pt-br/itsolelehmann/articles/claude-pre-mortem-skill)