soc-analyst
Run security operations center workflows — alert triage, investigation, escalation, and shift handover for detection and response.
- License 许可
- MIT
- 最近更新
- 2026-11-07
- Stars
- ★ 0
- Forks
- 1
01 前言 Front Matter
- name
- soc-analyst
- category
- security
- description
- Run security operations center workflows — alert triage, investigation, escalation, and shift handover for detection and response.
03 SKILL.md 正文
Overview
A SOC analyst turns the firehose of security telemetry into decisions: is this alert real, what is its scope, and what happens next? The work is triage under time pressure — distinguishing true positives from noise, containing what matters, and escalating with clean evidence. Good SOC work is a disciplined workflow, not heroics.
This skill covers alert triage, investigation method, severity classification, escalation, and the hygiene (runbooks, tuning, handover) that keeps a SOC effective instead of drowning.
The SOC is the organization's nervous system — it does not prevent every injury, but it determines how fast the body reacts. Analyst quality compounds: every well-investigated alert improves the runbook, every tuning ticket reduces tomorrow's noise, and every clean escalation builds IR trust. Protect analyst time for investigation, not just queue-clearing.
When to use
- Triaging SIEM/EDR/IDS alerts and deciding true vs false positive.
- Investigating a suspicious host, user, or network pattern end to end.
- Building or improving detection runbooks and alert tuning.
- Handling shift handover so nothing in flight gets dropped.
- Measuring and improving SOC performance (MTTD, MTTR, false-positive rate).
Core concepts
-
The triage question: "Is there evidence of malicious activity, and if so, what is the blast radius?" Everything else is detail.
-
Alert severity vs incident severity: an alert is a signal; an incident is a confirmed event with impact. Do not declare incidents from single alerts.
-
Kill chain / ATT&CK mapping: map observed activity to tactics (initial access → execution → persistence → exfil → impact). It tells you what to look for next and what the attacker still needs.
-
Blast radius first: before deep forensics, bound the scope — which hosts, users, data. Containment decisions need scope, not perfect attribution.
-
Evidence preservation: note timestamps, capture volatile data early, avoid destructive actions on the affected host until forensics is consulted.
-
Runbooks: every high-volume alert type gets a written triage path. Analysts follow the runbook; the runbook gets improved from analyst feedback.
-
Analyst tiers with clear escalation. L1 triages against runbooks, L2 investigates deeply, L3 hunts and engineers detections. Blurred tiers mean everything lands on whoever is most senior and available.
-
Threat intel with context. IOC feeds are noise without relevance filtering — prioritize intel matching your industry, geography, and technology stack.
-
Shift-left feedback. Analysts should have a direct, fast path to detection engineers — the people closest to the alerts know best what is broken.
Practical workflow
- Triage the queue: sort by severity and asset criticality. For each alert: check the runbook, validate the signal (is the source reliable? is the asset real?), and disposition within the SLA — true positive, false positive, or benign-true (real activity, not malicious).
- Investigate true positives: pivot on indicators — user, host, process tree, network connections, auth logs. Build a timeline. Ask: first observed, how it got in, what it touched, whether it persists, whether data left.
- Scope and classify: determine affected hosts/users/data; classify severity (use a fixed rubric: e.g., confirmed compromise of a production host = P1). Map to ATT&CK for the handover.
- Contain (with approval): isolate host, disable account, block indicator — per playbook authority levels. Record every containment action with a timestamp.
- Escalate cleanly: incident ticket with summary, timeline, scope, evidence links, actions taken, and open questions. Page incident response per the escalation matrix; do not freelance beyond your authority.
- Tune and hand over: false positives get a tuning ticket (not silent dismissal); shift handover lists in-flight investigations, their state, and next steps. Update the runbook with what you learned.
Triage checklist (per alert)
- [ ] Alert source and fidelity known; runbook exists
- [ ] Asset/user verified real and in scope
- [ ] Corroborating evidence checked (2+ sources before "true positive")
- [ ] Timeline started; first/last seen noted
- [ ] Blast radius estimated before containment
- [ ] Disposition + rationale logged; tuning ticket filed if FP
Sustaining the practice
- Review the top-10 noisiest rules monthly; tune or justify each
- Run quarterly purple-team or adversary-emulation validations of key detections
- Rotate analysts through threat-intel and detection-engineering stints to build skill
- Track analyst retention and burnout signals — turnover destroys institutional knowledge
Metrics that prove it works
- MTTD and MTTR, trended monthly by severity
- False-positive rate per detection rule (tune the worst offenders first)
- % of alerts triaged within SLA; escalation accuracy (incidents confirmed / escalations)
- Analyst feedback loop: tuning tickets filed and closed per quarter
Common pitfalls
- Alert fatigue → click-through triage. If analysts close alerts without investigation, the SOC is a checkbox. Tune noisy rules; measure FP rate per rule.
- Declaring incidents too early (or too late). Single uncorroborated alert ≠ incident; confirmed lateral movement you sat on for a shift = breach of duty. Use the rubric.
- Forensics-destroying "help." Rebooting, reimaging, or running AV sweeps before evidence capture can destroy the timeline. Contain first, preserve second, remediate third.
- No blast-radius estimate. Deep-diving one host while the attacker moves laterally elsewhere. Scope early, scope wide.
- Silent false positives. Every FP dismissed without a tuning ticket returns next week. The tuning backlog is a SOC health metric.
- Bad handovers. "Watch the queue" is not a handover. State, next steps, and watch items — written, every shift.
- Working the queue oldest-first. Triage by severity × asset criticality, not arrival order — a P1 aging while you clear informational alerts is a process failure.
- Not recording benign-true dispositions. "Real activity, not malicious" is tuning gold. Log it or the same alert returns forever.
- Treating every alert as equal effort. A 5-minute runbook triage and a 4-hour investigation are different work items. Staff and SLA them differently.
- Knowledge trapped in chat. Investigation findings shared only in chat threads are lost to the next shift. Conclusions belong in the ticket and the runbook.
04 同仓库更多 Skills
web-scrape
Turn web pages into LLM-ready markdown via the Firecrawl API (renders JS, anti-bot handling). Use when browser_open/fetch fails on JS-heavy pages, or when you need clean markdown/structured extraction from a site. Requires FIRECRAWL_API_KEY.
design-taste-frontend
tasteskill v2, the anti-slop frontend skill. Art direction for websites, distilled from a study of 247 curated gallery sites (One Page Love, Awwwards Sites of the Day). Deals researched directions (idea, composition, type, palette, photographs, section blueprints) from a library, enforces a strict no-slop house style with a real-browser inspector, and runs a look-and-fix loop on screenshots. Use this whenever you build, design, redesign, restyle, polish or review a landing page, marketing site, portfolio, product page, one-pager, hero or any public-facing web page, in any stack (HTML/CSS, React, Next.js, Vue, Svelte, Astro, Tailwind) — also when the user says a page looks generic, templated, "AI-made", sloppy, or asks to make it look designed, premium, award-level or less like every other site. Not for dashboards or app screens.
test-driven-development
Use when implementing any feature or bugfix, before writing implementation code
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes
requesting-code-review
Use when completing tasks, implementing major features, or before merging to verify work meets requirements
brainstorming
You MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation).