A zero-cost cognitive enhancement that enables AI agents to predict and prevent failures through adversarial pre-execution reasoning.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install premortem
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install premortem using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Premortem?
Premortem acts as a cognitive immune system for AI agents, integrating a sophisticated reasoning pattern directly into the agent's workflow. Instead of simply executing tasks, the agent uses this skill to imagine potential failure scenarios, identify root causes, and apply fixes before a single line of code is written or a command is run. By leveraging this addition to your Openclaw Skills collection, you ensure higher reliability, better code quality, and safer system interactions without adding API overhead or latency.
This is not a tool or an external API; it is a thinking pattern that triggers a cognitive reflex. It forces the agent to move beyond positive confirmation bias and adopt an adversarial perspective, identifying risks like hallucinated facts, off-by-one errors, or unintended side effects in shared systems before they manifest.
Premortem Use Cases
- Writing or refactoring complex, multi-file codebases to avoid regressions.
- Executing irreversible system commands or performing production deployments.
- Developing architectural recommendations where long-term trade-offs and scalability matter.
- Creating multi-step plans that require high alignment with user intent.
- Generating high-stakes content or research documentation for public distribution.
How Premortem Works
- Snapshot the Intent: The agent crystallizes the desired outcome into a single, clear success statement to ensure it hasn't drifted from the user's goal.
- Fast-Forward to Failure: The agent imagines the action has already failed and identifies the top three most likely failure modes using lenses such as correctness, security, and side effects.
- Inoculate: For every identified failure mode, the agent generates a specific patch or preventive measure to apply to its planned action.
- Execute with Confidence: The agent performs the final action, incorporating all generated patches for a cleaner, safer, and more accurate result.
Premortem Setup
Since Premortem is a cognitive protocol with zero dependencies, it is easily integrated into your agent's system prompt or configuration. To activate it within your Openclaw Skills environment, ensure the skill definition is loaded into the agent's active memory. No external libraries or API keys are required.
# Example of referencing the skill in an agent environment
openclaw install premortem
Premortem Data Schema & Taxonomy
Premortem utilizes a structured internal metadata taxonomy to process reasoning. While primarily internal, the logic follows this organizational schema:
| Component | Description |
|---|---|
| INTENT | A one-sentence summary of what the agent is trying to achieve. |
| FAILURE MODES | A prioritized list of risks categorized by Correctness, Security, or Side Effects. |
| PATCHES | Technical or logical fixes applied to prevent the identified failures. |
| STAKES LEVEL | Calibration data ranging from Low to Critical that dictates reasoning depth and time budget. |
Premortem Advanced Features
- Intensity Scaling: Dynamically adjusts reasoning depth from 1-second mental checks for trivial tasks to 10-second thorough protocols for critical system changes.
- The Inversion Principle: Switches from positive confirmation bias to negative adversarial thinking, effectively catching hidden bugs that standard reasoning misses.
- Silent Internal Reasoning: Operates as a background process to avoid cluttering the user interface while maintaining high-quality output.
- Domain-Specific Pattern Library: Includes specialized failure mode checklists for code generation, research, system actions, and architectural planning.
- Anti-Paralysis Guards: Built-in constraints to ensure reasoning never exceeds 10% of total action time or causes unnecessary delays.
SKILL.md
Loading
Related Openclaw Skills
Phantom Limb
Phantom Limb is a cognitive diagnostic tool that detects ghost references and silent dependencies that no longer exist in your live environment.
Peripheral Vision
A situational awareness skill that monitors upstream dependencies and downstream consumers for changes that could break your current work.
Pattern Mine
A structural analysis tool that identifies duplicated, inconsistent, and dead code patterns to guide intelligent refactoring.
Parallax
Parallax is a multi-perspective reasoning framework that evaluates technical decisions through five distinct stakeholder lenses to eliminate blind spots.
Proprioception
A self-spatial awareness system that gives AI agents a real-time sixth sense of their goal alignment, confidence boundaries, and output quality.
Reflex Arc
A zero-cost cognitive immune system that runs automatic pre-response checks to eliminate AI hallucinations and contradictions.
Featured*
