AGENT SKILL

zero-trust-architect

Design zero-trust architecture — identity-centric access, micro-segmentation, continuous verification, and phased migration.

License 许可
MIT
所属仓库
TanNhatCMS/jarvis
最近更新
2026-11-07
Stars
★ 0
Forks
1

01 前言 Front Matter

name
zero-trust-architect
category
security
description
Design zero-trust architecture — identity-centric access, micro-segmentation, continuous verification, and phased migration.

03 SKILL.md 正文

Overview

Zero trust is a strategy, not a product: never trust, always verify. Every access decision — user, device, workload — is authenticated, authorized, and encrypted based on identity and context, regardless of network location. It replaces the castle-and-moat model where "inside the network" meant trusted.

This skill covers the architecture: the pillars (identity, devices, networks, applications, data), policy design, and the phased migration path from legacy perimeter thinking. No vendor pitches — principles and practice.

Zero trust is a strategy for an uncomfortable truth: the network perimeter is gone, and 'inside' no longer means safe. The architecture replaces implicit trust with explicit, continuously evaluated decisions — which is powerful but operationally demanding. Organizations that succeed treat it as a multi-year program with visible milestones, not a product deployment with a go-live date.

When to use

  • Reducing blast radius so one compromised credential does not mean network-wide access.
  • Securing hybrid work: users and apps everywhere, no meaningful perimeter.
  • Planning phased migration away from VPN-centric remote access.
  • Meeting regulatory or customer expectations for modern access architecture.

Core concepts

  • The core tenets (NIST SP 800-207): all data sources and services are resources; all communication is secured; access is granted per-session; policy is dynamic (identity + device posture + behavior); the enterprise monitors integrity and security posture of all assets.

  • Identity as the perimeter: strong identity (phishing-resistant MFA), device posture, and risk signals drive each access decision — not source IP.

  • Least privilege per request: access is granted to specific resources for the session, continuously re-evaluated — not broad network segments for 8 hours.

  • Micro-segmentation: workloads isolated so lateral movement requires explicit authorization at each hop; assume breach between every segment.

  • Policy engine + policy enforcement: a central policy decision point evaluates trust signals; enforcement points (proxies, agents, gateways) carry it out. Design these before buying anything.

  • It is a journey: nobody "buys zero trust." Maturity progresses pillar by pillar; the roadmap matters more than the starting point.

  • Continuous diagnostics and mitigation (CDM). Posture signals (patch level, EDR health, encryption status) must flow into access decisions in near real time — stale signals mean stale trust.

  • Data-centric controls. Classification, labeling, and DLP follow the data beyond any perimeter; zero trust without data governance protects the pipes but not the water.

  • Legacy integration patterns. Mainframes, OT, and ancient apps will not do OIDC — plan compensating controls (gateways, jump hosts with session recording) rather than permanent exemptions.

Practical workflow

  1. Define protect surfaces: inventory crown jewels — the data, apps, and services that matter most. Zero trust starts by protecting what matters, not by boiling the ocean.
  2. Map transaction flows: for each protect surface, document who/what accesses it, from where, on which devices, and what the current implicit trust assumptions are.
  3. Architect the pillars: identity (phishing-resistant MFA, lifecycle), devices (posture/compliance checks), network (micro-segmentation, encrypted transport), applications (per-app access control), data (classification, DLP).
  4. Build policy: write access policies as code where possible — who (verified identity + group), what (specific resource), under which conditions (device compliant, risk low, business hours for sensitive). Start in monitor/audit mode.
  5. Migrate in phases: pilot on one protect surface (e.g., contractor access to one app via ZTNA); measure UX and security outcomes; expand surface by surface. Keep legacy paths only with explicit risk acceptance and sunset dates.
  6. Monitor and adapt: continuous diagnostics — policy decisions logged, anomalies investigated, posture signals fed back into risk scoring. Zero trust without monitoring is just complicated trust.

Phase plan template

  • Phase 0: inventory protect surfaces, map flows, baseline identity/device posture
  • Phase 1: identity hardening (phishing-resistant MFA, lifecycle) + pilot ZTNA on one surface
  • Phase 2: micro-segmentation around crown jewels; per-app policies in enforce mode
  • Phase 3: expand surfaces; retire legacy VPN paths; continuous verification everywhere
  • Ongoing: policy-as-code reviews, posture-signal tuning, maturity reassessment

Sustaining the practice

  • Reassess maturity annually against a model (e.g., CISA ZTMM) and publish the delta
  • Review policy-as-code changes with the same rigor as application code
  • Measure user friction continuously — degraded UX drives shadow IT
  • Sunset legacy access paths on committed dates, not 'when convenient'

Metrics that prove it works

  • Protect-surface coverage % under zero-trust policy
  • % of access flows via ZTNA/identity-aware paths vs legacy network paths
  • Policy-decision audit completeness (can you reconstruct who accessed what?)
  • Phishing-resistant authentication % across the estate

Common pitfalls

  • Buying "zero trust in a box." It is an architecture of identity, segmentation, and policy — no single product delivers it.
  • Ignoring identity fundamentals. Zero trust on top of weak passwords and no MFA is theater. Identity hygiene is phase one.
  • Big-bang migration. Ripping out network access wholesale breaks the business. Phase by protect surface.
  • Forgetting devices and data. Identity-only zero trust misses compromised endpoints and unclassified data. All pillars matter.
  • Policies nobody can understand. Over-complex policy-as-code becomes unmaintainable. Keep policies readable, reviewed, and tested.
  • No UX measurement. If the new model makes work painful, users route around it (shadow IT). Measure and optimize the experience, not just the security.
  • Zero-trust slideware. Architecture diagrams without defined protect surfaces and mapped transaction flows are marketing. Start with the crown jewels inventory.
  • Forgetting non-human identities. Service accounts and workloads need the same verify-every-request treatment, or they become the bypass.
  • Declaring victory after SSO + MFA. Identity hardening is phase one of five pillars, not the destination. Devices, networks, apps, and data need equal attention.
  • Policy exceptions without owners. Every legacy bypass needs a named owner, compensating controls, and a sunset date — or it becomes the permanent architecture.

04 同仓库更多 Skills

TanNhatCMS
TanNhatCMS/jarvis

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.

★ 0
TanNhatCMS
TanNhatCMS/jarvis

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.

★ 0
TanNhatCMS
TanNhatCMS/jarvis

test-driven-development

Use when implementing any feature or bugfix, before writing implementation code

★ 0
TanNhatCMS
TanNhatCMS/jarvis

systematic-debugging

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes

★ 0
TanNhatCMS
TanNhatCMS/jarvis

requesting-code-review

Use when completing tasks, implementing major features, or before merging to verify work meets requirements

★ 0
TanNhatCMS
TanNhatCMS/jarvis

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).

★ 0