Premortem

Cognitive immune system for AI agents. Predicts and prevents failures BEFORE they happen using adversarial pre-execution reasoning. Zero dependencies, zero c...

This skill appears to implement a legitimate internal 'premortem' thinking pattern and needs no credentials or installs, but the always:true setting plus the explicit instruction to keep premortems internal reduces transparency. Before installing, consider: 1) Do you want a skill forced into every agent run? If not, ask for always:false or an opt-in trigger. 2) Ask the author to log or surface a short, auditable summary when a premortem produces a 'critical risk' or a patch that changes behavior, so you can review decisions. 3) Confirm provenance (the SKILL.md lists a GitHub URL but registry metadata lacks a homepage) so you can inspect source history. 4) If you require full traceability for compliance or debugging, avoid enabling always:true or require the skill to reveal premortem summaries by default. If you trust the author and prefer the quality improvement, installing with the ability to opt out or audit premortem summaries is a safer choice.

Type: OpenClaw Skill Name: premortem Version: 1.0.0 The OpenClaw skill bundle 'premortem' is designed to enhance an AI agent's reasoning by instructing it to perform a 'premortem' analysis before taking actions. This involves identifying potential failure modes (including security vulnerabilities) and patching them proactively. All files (SKILL.md, _meta.json, references/benchmarks.md, references/cognitive-science.md) consistently describe this self-correction mechanism. There is no evidence of malicious intent, data exfiltration, unauthorized command execution, persistence mechanisms, or prompt injection designed to subvert the agent for harmful purposes. The instructions related to 'security' are for the agent to prevent vulnerabilities in its own output, not to create or exploit them.