# Agent Skill Safety Review — Visual Preview Pack

Share-ready visual/caption pack for distributing the Agent Skill Safety Review funnel outside X.

Main funnel: https://jarvislandingdeploy.vercel.app/agent-skill-safety-review.html
Sample report: https://jarvislandingdeploy.vercel.app/agent-skill-safety-sample-report.html
Scorecard: https://jarvislandingdeploy.vercel.app/agent-skill-safety-scorecard.html
30-minute kit: https://jarvislandingdeploy.vercel.app/agent-skill-review-30-minute-kit.html
Directory pack: https://jarvislandingdeploy.vercel.app/agent-skill-safety-directory-submission-pack.md

## Preview cards

- Overview card: https://jarvislandingdeploy.vercel.app/media/agent-skill-safety-preview-01.svg
- Review checks card: https://jarvislandingdeploy.vercel.app/media/agent-skill-safety-preview-02.svg
- Sample report card: https://jarvislandingdeploy.vercel.app/media/agent-skill-safety-preview-03.svg
- 30-minute kit card: https://jarvislandingdeploy.vercel.app/media/agent-skill-safety-preview-04.svg

## Recommended screenshot order

1. Main offer page hero: value prop + scorecard / kit / sample report / $29 mini-report CTA.
2. Browser-local scorecard: 100-point rubric, no data stored, immediate verdict.
3. Sample mini-report: 63/100 verdict, top risks, receipts, evals, rollback gaps.
4. 30-minute kit: permission matrix, trust-boundary checks, copy/paste verdict.

## Short caption

Agent skills are becoming software packages. Before installing, selling, or delegating one, review its tool authority, memory policy, receipts, rollback, evals, and failure behavior.

Free scorecard + 30-minute kit + sample mini-report:
https://jarvislandingdeploy.vercel.app/agent-skill-safety-review.html

## Community caption

I made a practical review pack for portable AI-agent skills. The premise is simple: a skill is not just a prompt; it is delegated authority. The checklist focuses on the boring parts that prevent production failures: allowed tools, forbidden actions, memory writes, data boundaries, receipt logs, exception behavior, rollback, and eval cases.

Useful if you are publishing, buying, installing, or delegating skills across OpenClaw / Claude Code / Hermes / Codex-style agents.

Start here: https://jarvislandingdeploy.vercel.app/agent-skill-safety-review.html

## Warm reply caption

This is exactly the trust gap I keep seeing with reusable agent skills. I put together a free scorecard + sample report for reviewing tool authority, memory blast radius, receipts, evals, and rollback before a skill crosses a real boundary:
https://jarvislandingdeploy.vercel.app/agent-skill-safety-review.html

## Anti-spam rule

Do not drop these links cold into unrelated threads. Use this only when someone is already discussing agent skills, marketplaces, Claude Code/OpenClaw skills, tool permissions, prompt injection, skill scanners, agent memory, or production-agent safety.