AgentConn

Anthropic Cybersecurity Skills

Framework Agnostic Intermediate Security Open Source

mukul975/Anthropic-Cybersecurity-Skills is a community-built skill bundle that publishes 754 structured cybersecurity skills as agentskills.io-standard SKILL.md instruction packs. Each skill is mapped against five canonical frameworks — MITRE ATT&CK, NIST CSF 2.0, MITRE ATLAS, D3FEND, and NIST AI RMF — providing direct traceability from a runtime action to a recognized control or technique. The bundle works across Claude Code, GitHub Copilot, Codex, Cursor, and other agent harnesses that load skills at runtime. The repo entered GitHub trending on May 25 2026 with +999 stars in a 24-hour window (9K stars total), riding the convergence of (a) Pope Leo XIV's Magnifica Humanitas encyclical on AI governance the same day and (b) Microsoft's release of the agent-governance-toolkit covering 10/10 OWASP Agentic Top 10. It is one of the four Skills-targeted repos on the top trending board today, alongside multica-ai/andrej-karpathy-skills, affaan-m/ECC, and multica-ai/multica.

Input / Output

Accepts

skill-name framework-mapping

Produces

installed-skill control-traced-action

Overview

mukul975/Anthropic-Cybersecurity-Skills is the largest community-built cybersecurity skill bundle for AI agents to date. 754 SKILL.md packs, each cross-referenced to canonical control frameworks, distributed as a single repository that any skill-loading harness can install.

Why It Matters Now

The repo crossed 9K stars on the same day Pope Leo XIV’s Magnifica Humanitas encyclical landed and Microsoft’s agent-governance-toolkit sat alongside it on the trending board. The market signal is that every skill-loading harness needs a security-skills bundle within reach, and this one has the framework mappings the compliance auditors will want.

Framework Coverage

  • MITRE ATT&CK — every offensive-technique skill traceable to a TTP.
  • NIST CSF 2.0 — defensive-posture skills aligned to Identify/Protect/Detect/Respond/Recover.
  • MITRE ATLAS — adversarial ML threat library.
  • D3FEND — defensive countermeasure ontology.
  • NIST AI RMF — AI-specific risk-management mappings.

Use Cases

Security operations teams use it as the runtime skill library for incident-response agents (Detect/Respond TTPs), purple-team exercise scripting (ATT&CK + D3FEND pairs), and EU AI Act / Colorado AI Act compliance evidence collection (NIST AI RMF mappings). The August 2026 EU AI Act high-risk obligations and June 2026 Colorado deadline make the framework traceability immediately load-bearing.

Compatibility

Loads in any harness that supports the agentskills.io SKILL.md standard — Claude Code, GitHub Copilot, OpenAI Codex CLI, Cursor, and others. Pair with the Microsoft Agent Governance Toolkit for OWASP Agentic Top 10 runtime enforcement on top of the skill traceability.

Tags

#skills #cybersecurity #mitre-attack #nist #d3fend #owasp-agentic #ai-governance #claude-skills-pattern #2026

Compatible Agents

AI agents that work well with Anthropic Cybersecurity Skills.

Similar Skills