How to use the Safety Alert Deck agent skill

What the skill makes the agent do

An agent skill is a folder with a SKILL.md file that gives an AI agent extra instructions for a particular kind of task. This one covers presentation slides for safety, risk, incident post-mortems, red-team work and policy-as-code topics. When it applies, the agent builds the deck in a warning-style visual layout rather than a default one.

• Diagonal 45-degree red and black hazard stripes at the top and bottom of slides
• Red strikethrough headings for negative or rejected points
• Green, amber and red tier cards labelled L1, L2 and L3
• Round-dot status alert boxes
• Code blocks styled for policy-as-code YAML, with a red left border and highlighted "bad" keywords
• Red and green checklists and a stacked bar chart for incidents

Install the skill from the ZIP download

Download the deck-safety-alert ZIP from this page and unzip it. The package contains two files: SKILL.md and example.html. Place the unzipped deck-safety-alert folder in one of two locations, depending on how you want to use it.

For a personal setup available across your projects, put it in ~/.claude/skills/. For a single project, put it in that project's .claude/skills/ folder. The skill is then picked up by agents working in that location.

Alternatively, if you use the Claude apps, you can upload the skill under Settings → Capabilities → Skills instead of copying files by hand.

• Personal install: ~/.claude/skills/deck-safety-alert/
• Project install: <your-project>/.claude/skills/deck-safety-alert/
• Claude apps: Settings → Capabilities → Skills, then upload

Trigger the skill with a natural request

You do not need a special command. Describe a deck that fits the skill's purpose and mention the kind of content: safety, security, risk, incident, red team or policy-as-code. The agent uses the skill when the request matches its description.

Be specific about the content. Naming the topic and the audience usually gives a better deck than a generic request.

• "Make a 10-slide incident post-mortem deck for last week's database outage."
• "Create a slide deck presenting our red-team results for the payments API, with L1 to L3 severity tiers."
• "Build a policy-as-code review deck showing the rules we block in CI, with the bad patterns highlighted."

Check the output and adapt it

Open the example.html file included in the package to see the finished look before you ask for a deck. It shows how the stripes, tier cards and alert boxes are meant to appear together.

Review the generated slides for accuracy first. The skill controls the visual style; the facts, severity labels and wording still come from your request and your source material.

• Compare your output against example.html to confirm the style is what you expected
• Check every tier label and incident detail before presenting

Requirements and safety notes

The skill lists no requirements. It does not run scripts and does not use the network, according to its listing. It consists of instructions and an example HTML file.

• Requires: nothing listed
• Runs scripts: no
• Uses network: no
• Read the SKILL.md before installing any skill you have not written yourself, since an agent will follow its instructions.