# Twitter/X Launch Thread

Post as a thread. Pin the first tweet. Include the repo link in tweet 1 and tweet 7.

---

## Tweet 1 (268 characters)

I just open-sourced 611 cybersecurity skills for AI agents.

From malware analysis with Volatility to cloud pentesting with Pacu -- structured so Claude Code, Copilot, and any AI agent can use them.

MIT licensed. All 24 subdomains of cybersecurity.

github.com/mukul975/Anthropic-Cybersecurity-Skills

---

## Tweet 2 (277 characters)

The problem: AI agents are great at coding but terrible at cybersecurity.

Ask Claude to analyze a memory dump and you get generic advice.

Give it a structured skill with the exact Volatility plugin sequence, and it gives you the precise commands a senior analyst would use.

---

## Tweet 3 (270 characters)

Each skill uses progressive disclosure:

YAML frontmatter = WHEN to activate (triggers, domain, tags)
Markdown body = HOW to execute (exact commands, decision trees, validation)

The agent loads the frontmatter for routing, then the full body only when it needs the details.

---

## Tweet 4 (280 characters)

611 skills across 24 subdomains:

- Cloud Security (48)
- Threat Intelligence (43)
- Web App Security (41)
- Threat Hunting (35)
- Malware Analysis (34)
- Digital Forensics (34)
- SOC Operations (33)
- Network Security (33)
- IAM (33)
- OT/ICS Security (28)
- And 14 more

---

## Tweet 5 (257 characters)

These aren't generic cheat sheets. Every skill has:

- Real tool commands (not "use a scanner")
- MITRE ATT&CK technique IDs
- NIST/CIS benchmark references
- Decision trees for edge cases
- Practitioner helper scripts
- Filled-in report templates

---

## Tweet 6 (243 characters)

Why this matters for the security industry:

AI agents will increasingly assist with security work. The question isn't IF but HOW WELL.

Structured skill databases are how we go from "vaguely helpful AI" to "AI that knows the right Splunk query for T1059.001."

---

## Tweet 7 (248 characters)

The repo is MIT licensed and follows the agentskills.io open standard.

Looking for contributors -- especially practitioners who want to encode their expertise for AI agents.

If you write runbooks, you can write skills.

github.com/mukul975/Anthropic-Cybersecurity-Skills