a2a-security-review
Review agent-to-agent (A2A) / multi-agent-system trust: peer identity and authentication, message integrity, capability-negotiation trust, and delegation-chain…
Rank a set of enumerated threats or findings by risk (likelihood × impact) and map each to a prioritized mitigation. Use after STRIDE/PASTA enumeration or any time you have a threat/finding list that needs prioritization.
$ npx -y skills add jassics/awesome-claude-security --skill risk-rank --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/risk-rankContext preview
The summary Claude sees to decide when to auto-load this skill.
Rank a set of enumerated threats or findings by risk (likelihood × impact) and map each to a prioritized mitigation. Use after STRIDE/PASTA enumeration or any time you have a threat/finding list that needs prioritization.
name: risk-rank description: >- Rank a set of enumerated threats or findings by risk (likelihood × impact) and map each to a prioritized mitigation. Use after STRIDE/PASTA enumeration or any time you have a threat/finding list that needs prioritization.
A defensible risk ranking that drives remediation order, with each item tied to a mitigation and a suggested priority.
1. **Score likelihood** (1–5): attacker capability required, exposure/reachability, exploit availability, existing controls. 2. **Score impact** (1–5): confidentiality/integrity/availability damage, blast radius, business/compliance consequence. 3. **Risk = likelihood × impact** (1–25). Banding: 1–5 Low · 6–11 Medium · 12–18 High · 19–25 Critical. (For vuln-level items, prefer CVSS via `security-reporting:cvss` and reconcile the two.) 4. **Map mitigation** per item: the control that most reduces likelihood or impact; note whether it already exists (gap vs. present). 5. **Prioritize** by risk, then by remediation cost/effort (quick high-risk wins first).
A ranked table: ID · threat/finding · likelihood · impact · risk · band · mitigation · effort · priority. Plus a short "do first / do next / accept" list.
State the scoring assumptions so the ranking is reproducible and challengeable. Keep likelihood and impact independent — don't double-count exposure in both.
A Claude Code plugin marketplace for the full cybersecurity & GenAI-security lifecycle — from recon and threat modeling to detection engineering, GRC, and CISO-level strategy. A pentester knows which OWASP test bends a broken-access-control endpoint.
Repo: jassics/awesome-claude-security
Review agent-to-agent (A2A) / multi-agent-system trust: peer identity and authentication, message integrity, capability-negotiation trust, and delegation-chain…
Test the agent execution harness/runtime itself — LangChain/LangGraph, AutoGen, CrewAI, custom ReAct-style loops, or computer-use/browser-use agents — for…
Assess an autonomous / tool-using AI agent for security end-to-end: tool privileges, autonomy and approval boundaries, excessive agency, memory/state…
Test what an AI agent will actually do without human confirmation, including under injected-goal / prompt-injection scenarios, to validate its autonomy and…
Review the security of MCP (Model Context Protocol) servers/clients an agent uses: server trust tier, tool/resource description and result poisoning,…
Inventory the tools/functions an AI agent can call and audit their privileges, side effects, and approval requirements to find excessive-agency and…