a2a-security-review
Review agent-to-agent (A2A) / multi-agent-system trust: peer identity and authentication, message integrity, capability-negotiation trust, and delegation-chain…
Triage raw vulnerability-scanner output (Nessus/Qualys/Trivy/Grype/OpenVAS, cloud or container scans): normalize, deduplicate, validate, and cut false positives so the list is trustworthy before prioritization. Use when handed a scan export or a pile of findings that needs to
$ npx -y skills add jassics/awesome-claude-security --skill vulnerability-scan-triage --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/vulnerability-scan-triageContext preview
The summary Claude sees to decide when to auto-load this skill.
Triage raw vulnerability-scanner output (Nessus/Qualys/Trivy/Grype/OpenVAS, cloud or container scans): normalize, deduplicate, validate, and cut false positives so the list is trustworthy before prioritization. Use when handed a scan export or a pile of findings that needs to
name: vulnerability-scan-triage description: >- Triage raw vulnerability-scanner output (Nessus/Qualys/Trivy/Grype/OpenVAS, cloud or container scans): normalize, deduplicate, validate, and cut false positives so the list is trustworthy before prioritization. Use when handed a scan export or a pile of findings that needs to become an actionable backlog.
A clean, deduplicated, validated finding set — false positives removed and each remaining item tied to a real affected asset — ready to feed `vulnerability-prioritization`.
1. **Normalize** — parse the export(s) into a common shape: CVE/identifier, title, affected asset(s), installed vs. fixed version, scanner severity, plugin/check ID, first-seen date. Merge multiple scanners into one inventory. 2. **Deduplicate** — collapse the same vuln on the same asset reported by multiple scanners or scans; group "same CVE across N assets" so you triage the vuln once, not N times. 3. **Validate** — separate confirmed from suspected. Check for known false-positive patterns: backported fixes (distro patched without bumping the version string), not-installed/not-loaded code paths, mitigations already in place, or a check that only inspects a banner. Mark each as confirmed / needs-verification / FP. 4. **Enrich for context** — attach asset criticality, environment (prod/non-prod), internet exposure, and data sensitivity from the CMDB/asset inventory. This is what later drives priority. 5. **Hand off** — produce the cleaned set and route to `vulnerability-prioritization`.
A normalized finding table — identifier · affected assets · installed/fixed version · status (confirmed/FP/needs-verification) · scanner(s) · asset context — plus a short note on what was deduped and which FP patterns were applied. Use `security-reporting` for any formal writeup.
Garbage in, garbage out: prioritization is only as good as the triage. Backported patches are the most common false positive on Linux distros — verify against the distro's security tracker, not just the version number. Track first-seen dates so SLA clocks start correctly.
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…