# mukul975/Anthropic-Cybersecurity-Skills

Source: [GitHub](https://github.com/mukul975/Anthropic-Cybersecurity-Skills)  
Feed7 permalink: https://feed7.dev/p/anthropic-cybersecurity-skills-1k6d1bw  
Published: Unknown  
Trust: Needs Review (needs_review)

## Why Included

This community library packages 817 security workflows as agent skills, offering broad framework-mapped guidance while requiring careful review for offensive and dual-use tasks.

## Source Summary

The repository contains **817 skills** across **29 security domains**, using the agentskills.io format. Coverage includes 805 ATT&CK mappings, 804 NIST CSF mappings, and smaller sets for D3FEND, AI RMF, F3, and ATLAS.

## Practical Implication

Treat it as a searchable playbook layer for security agents: frontmatter supports cheap discovery, while full skill files provide prerequisites, procedures, decision points, and verification steps. Review the selected workflow and required access before execution.

## Agent-Ready Context

The repository contains **817 skills** across **29 security domains**, using the agentskills.io format. Coverage includes 805 ATT&CK mappings, 804 NIST CSF mappings, and smaller sets for D3FEND, AI RMF, F3, and ATLAS.

Treat it as a searchable playbook layer for security agents: frontmatter supports cheap discovery, while full skill files provide prerequisites, procedures, decision points, and verification steps. Review the selected workflow and required access before execution.

This is an **independent community project**, not an Anthropic release. It includes offensive and dual-use material, so use must stay within owned systems or explicit written authorization; the repository’s quality claims are not independently validated here.

## Connected Context

Feed7 judgment across 479 accumulated Signals:

This expands the skill-catalog pattern into a security-specific playbook layer with unusually broad framework mapping and discovery metadata. Against prior candidates, its scale strengthens the case for selective routing, authorization checks, and supply-chain review rather than wholesale installation. The mappings improve findability, but neither they nor the repository’s own quality claims establish workflow correctness or safe execution.

- [zhaoxuya520/reverse-skill](https://feed7.dev/p/reverse-skill-1e4jlfw) — reverse-skill supplies a narrower authorization-first, evidence-linked execution pattern that could govern selection and use of this much broader security playbook catalog.
- [We Vetted 2000 AI Skills Before They Reached Developers — Lucas Palma, Nubank](https://feed7.dev/p/we-vetted-2000-ai-skills-before-they-reached-developers-lucas-palma-nuba-0k4ehkz) — The catalog’s 817 independent, dual-use skills make Nubank’s deterministic scanning, LLM review, and distribution gating directly relevant as prerequisites for trusted adoption.
- [affaan-m/ECC](https://feed7.dev/p/ecc-0438yju) — Both expose large cross-domain skill surfaces whose breadth increases audit and maintenance work, reinforcing selective installation and verification rather than assuming more skills improve the harness.
- [In Code They Act, In Proof We Trust — Erik Meijer, Leibniz Labs](https://feed7.dev/p/in-code-they-act-in-proof-we-trust-erik-meijer-leibniz-labs-1lyno2y) — Machine-checkable approval before side effects would strengthen the repository’s stated authorization boundary, which currently depends on users reviewing workflows and access before execution.

## Context Map

- Layer: agent
- Domains: security
- Topics: skills, tool-use

## Uncertainty

- This is an **independent community project**, not an Anthropic release. It includes offensive and dual-use material, so use must stay within owned systems or explicit written authorization; the repository’s quality claims are not independently validated here.

## Agent Instruction

Use this item as source-backed context. Do not invent claims beyond the linked source. If this item conflicts with another source, call out the conflict.