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 a SIEM/EDR alert end-to-end — validate it's real, enrich and scope it, reach a consistent verdict, and decide escalate vs. close with documented rationale. Use when working a SOC alert queue and you need a repeatable, defensible triage.
$ npx -y skills add jassics/awesome-claude-security --skill alert-triage --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/alert-triageContext preview
The summary Claude sees to decide when to auto-load this skill.
Triage a SIEM/EDR alert end-to-end — validate it's real, enrich and scope it, reach a consistent verdict, and decide escalate vs. close with documented rationale. Use when working a SOC alert queue and you need a repeatable, defensible triage.
name: alert-triage description: >- Triage a SIEM/EDR alert end-to-end — validate it's real, enrich and scope it, reach a consistent verdict, and decide escalate vs. close with documented rationale. Use when working a SOC alert queue and you need a repeatable, defensible triage.
A consistent, evidence-backed verdict on an alert — true positive / false positive / benign-true-positive — with the right next action (escalate, close, tune) and a record that the next analyst can follow.
1. **Understand the alert** — what detection fired, the underlying logic, the ATT&CK technique, and why it triggered. Read the rule, don't guess. 2. **Validate** — is this real activity or a known false-positive pattern? Confirm against the source telemetry, not just the alert summary. 3. **Enrich** — add context: user/asset criticality, the involved IOCs (`threat-intelligence:ioc-enrichment`), process lineage, recent related alerts, and whether it matches known-good behavior. 4. **Scope** — pivot to see if it's isolated or part of a broader pattern (same host/ user/campaign, lateral movement). Check for related alerts you should correlate. 5. **Decide & act:**
evidence package and scope assembled.
`detection-engineering:detection-rule-development` (don't just mute). 6. **Record** — verdict, evidence, actions, and timestamps.
A triage record: alert · ATT&CK technique · verdict · evidence · scope · action (escalate/close/tune) · rationale. Confirmed incidents → `dfir`; recurring FPs → `detection-engineering`.
Consistency is the goal — the same alert should get the same verdict regardless of analyst. Always validate against source telemetry before deciding, and feed recurring false positives back into detection tuning rather than muting them (muting hides the problem and erodes coverage). Preserve evidence and rationale for every verdict.
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…