Postmortem

What Is Postmortem?

The Postmortem skill for Claude Code is a powerful tool designed to facilitate honest, constructive analysis of failures in development and organizational processes. Accessible through the /em:postmortem <event> command, this skill steers teams away from unproductive blame games and empty platitudes, focusing instead on rigorous investigation and deep understanding. The goal is not to assign fault, but to uncover the true, systemic causes behind missed objectives—whether that’s a failed product launch, a missed revenue target, or a hiring misfire. By systematically dissecting what happened, why it occurred, and what should change as a result, the Postmortem skill aims to turn setbacks into valuable learning opportunities that drive continuous improvement.

Why Use Postmortem?

Postmortem analysis is a critical component of high-functioning development and executive teams. Traditional postmortems often devolve into two unhelpful extremes: the blame session or the whitewash. In the former, individuals are scapegoated, stifling honest conversation and ensuring a recurrence of the same issue in a new guise. In the latter, teams pay lip service to learning by generating vague action items without addressing root causes, resulting in no real change.

The Postmortem skill provides a structured, neutral framework to avoid these pitfalls. It encourages teams to ask, “What conditions made this outcome predictable in hindsight?” rather than “Whose fault was it?” This approach maximizes learning, prevents recurrence, and improves long-term system robustness. In the high-stakes world of software development and business operations, a well-conducted postmortem is a proven way to boost resilience, foster openness, and accelerate growth.

How to Get Started

Using the Postmortem skill is straightforward. After a critical event, such as a failed release or missed target, initiate the analysis with the following command in Claude Code:

/em:postmortem <event>

Where <event> is a concise identifier for the incident you want to analyze. For example:

/em:postmortem Q3 ARR shortfall

The skill will then prompt you to define the event precisely and guide you through a sequence of reflective questions. Here’s a practical code example for a more detailed invocation:

/em:postmortem "Feature X launch failure - delayed delivery and low adoption"

The output will structure your inquiry, asking for specifics such as:

• What was the expected outcome?
• What was the actual outcome?
• When did the deviation first become apparent?
• What were the financial, operational, and reputational impacts?

This structured approach ensures that key details are captured and that the subsequent analysis is both actionable and objective.

Key Features

The Postmortem skill is designed around a clear, step-by-step framework that extracts maximum learning from each failure:

1. Precise Event Definition: The skill prompts for a factual, detailed statement of the event. Ambiguity is avoided in favor of measurable outcomes, timelines, and impact assessments.
2. Root Cause Analysis: Instead of focusing on individual missteps, the skill leads users to examine the broader conditions and systemic factors that contributed to the failure.
3. Sequence of Events: The tool guides you to reconstruct the timeline, pinpointing when and how the gap between expectation and reality emerged.
4. Impact Assessment: Users are encouraged to quantify the effects—financial, operational, and reputational—of the event.
5. Actionable Learning: The process culminates in identifying concrete changes to prevent recurrence, ensuring that lessons are translated into improved practices.

Example output snippet:

## Event Definition
- Expected: Launch Feature X to 10,000 users by June 1
- Actual: Launched to 2,500 users on July 10, with adoption at 10% of forecast

## Root Causes
- Over-optimistic delivery estimates
- Unclear requirements leading to rework
- Insufficient user onboarding resources

## Impact
- Lost $50K in projected revenue
- Negative customer feedback on delays

## Actions
- Establish buffer in delivery forecasts
- Formalize requirements gathering stage
- Allocate dedicated onboarding team

Best Practices

• Be Specific: Vague statements hinder learning. Quantify outcomes and impacts wherever possible.
• Maintain Psychological Safety: Ensure all participants understand the focus is on systems and processes, not personal blame.
• Follow the Framework: Allow the skill to guide you through each step. Skipping steps can result in incomplete or biased analysis.
• Document and Share: Make postmortem reports available to relevant stakeholders to maximize organizational learning.
• Review Regularly: Periodically revisit past postmortems to check if recommended changes have been implemented and effective.

Important Notes

• The Postmortem skill is not a substitute for accountability, but it reframes accountability in terms of system and process improvement rather than individual fault.
• To maximize effectiveness, postmortems should be conducted promptly after an event while details are still fresh.
• The skill is most effective when integrated into regular team routines and used consistently, not just after major failures.
• As with any analysis tool, the quality of insights depends on the honesty and rigor of the participants.
• The source code and further documentation are available on GitHub for teams looking to customize or extend the skill’s functionality.