本文へ移動
cccskills
無料GitHub で公開

configuring-identity-aware-proxy-with-google-iap

Configures Google Cloud Identity-Aware Proxy (IAP) via gcloud to enforce per-request identity verification on Compute Engine, App Engine, Cloud Run, and GKE, including IAM bindings, Access Context Manager access levels, session/reauth settings, and service-account programmatic access. Use when replacing VPN access with identity-based access to GCP backends or configuring context-aware, zero-trust policies for Google Cloud services.

インストール方法を見る

含まれるファイル(8)

  • SKILL.md14.4 KB
  • assets/template.md2.2 KB
  • LICENSE11.0 KB
  • references/api-reference.md1.6 KB
  • references/standards.md1.7 KB
  • references/workflows.md2.4 KB
  • scripts/agent.py4.4 KB
  • scripts/process.py7.7 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Configuring Identity-Aware Proxy with Google IAP

When to Use

  • When protecting Google Cloud applications (App Engine, Cloud Run, GKE, Compute Engine) with identity-based access
  • When implementing context-aware access requiring device posture and location verification
  • When providing secure access to internal tools without VPN or public IP exposure
  • When needing per-request authentication and authorization for web applications and TCP services
  • When configuring programmatic access to IAP-protected resources using service accounts

Do not use for non-HTTP applications that cannot be placed behind an HTTPS load balancer, for public-facing applications that need unauthenticated access, or when applications handle their own authentication and IAP would conflict with existing auth flows.

Prerequisites

  • Google Cloud project with billing enabled
  • IAP API enabled (gcloud services enable iap.googleapis.com)
  • Application deployed behind HTTPS Load Balancer, App Engine, or Cloud Run
  • Cloud Identity or Google Workspace for user management
  • Access Context Manager API enabled for access levels
  • OAuth consent screen configured for the project

Workflow

Step 1: Enable IAP on Backend Services

Configure IAP for different GCP compute platforms.

# Enable required APIs
gcloud services enable iap.googleapis.com
gcloud services enable accesscontextmanager.googleapis.com

# Create OAuth consent screen
gcloud iap oauth-brands create \
  --application_title="Internal Applications" \
  --support_email=security@company.com

# Create OAuth client
gcloud iap oauth-clients create \
  projects/PROJECT_ID/brands/BRAND_ID \
  --display_name="IAP Web Client"

# === Enable IAP on Compute Engine Backend Service ===
gcloud compute backend-services update my-backend-service \
  --iap=enabled,oauth2-client-id=CLIENT_ID,oauth2-client-secret=CLIENT_SECRET \
  --global

# === Enable IAP on App Engine ===
gcloud iap web enable \
  --resource-type=app-engine \
  --oauth2-client-id=CLIENT_ID \
  --oauth2-client-secret=CLIENT_SECRET

# === Enable IAP on Cloud Run ===
# First grant IAP service account the Cloud Run Invoker role
gcloud run services add-iam-policy-binding my-service \
  --member="serviceAccount:service-PROJECT_NUM@gcp-sa-iap.iam.gserviceaccount.com" \
  --role="roles/run.invoker" \
  --region=us-central1

# Enable IAP on the Cloud Run backend service
gcloud compute backend-services update my-cloud-run-backend \
  --iap=enabled,oauth2-client-id=CLIENT_ID,oauth2-client-secret=CLIENT_SECRET \
  --global

# === Enable IAP TCP Forwarding for SSH/RDP ===
# No load balancer needed - uses IAP tunnel
gcloud compute instances add-iam-policy-binding my-vm \
  --member="group:developers@company.com" \
  --role="roles/iap.tunnelResourceAccessor" \
  --zone=us-central1-a

# SSH through IAP tunnel
gcloud compute ssh my-vm --zone=us-central1-a --tunnel-through-iap

# RDP through IAP tunnel
gcloud compute start-iap-tunnel my-windows-vm 3389 \
  --local-host-port=localhost:3390 \
  --zone=us-central1-a

Step 2: Configure IAM Bindings for Access Control

Grant access to specific users and groups with optional access level conditions.

# Grant basic access to a group
gcloud iap web add-iam-policy-binding \
  --resource-type=backend-services \
  --service=my-backend-service \
  --member="group:engineering@company.com" \
  --role="roles/iap.httpsResourceAccessor"

# Grant access with access level condition
gcloud iap web add-iam-policy-binding \
  --resource-type=backend-services \
  --service=finance-app \
  --member="group:finance@company.com" \
  --role="roles/iap.httpsResourceAccessor" \
  --condition='expression=request.auth.access_levels.exists(x, x == "accessPolicies/POLICY_ID/accessLevels/corporate-device"),title=RequireCorporateDevice,description=Requires managed corporate device'

# Grant access only during business hours
gcloud iap web add-iam-policy-binding \
  --resource-type=backend-services \
  --service=admin-console \
  --member="group:admins@company.com" \
  --role="roles/iap.httpsResourceAccessor" \
  --condition='expression=request.time.getHours("America/New_York") >= 8 && request.time.getHours("America/New_York") <= 18 && request.time.getDayOfWeek("America/New_York") >= 1 && request.time.getDayOfWeek("America/New_York") <= 5,title=BusinessHoursOnly'

# Grant access to a specific URL path
gcloud iap web add-iam-policy-binding \
  --resource-type=backend-services \
  --service=internal-api \
  --member="group:api-consumers@company.com" \
  --role="roles/iap.httpsResourceAccessor" \
  --condition='expression=request.path.startsWith("/api/v2/"),title=APIv2Access'

Step 3: Create Access Levels with Access Context Manager

Define context-based access requirements using device attributes and network conditions.

# Create access level requiring encrypted corporate device
cat > managed-device.yaml << 'EOF'
- devicePolicy:
    allowedEncryptionStatuses:
      - ENCRYPTED
    osConstraints:
      - osType: DESKTOP_WINDOWS
        minimumVersion: "10.0.19045"
      - osType: DESKTOP_MAC
        minimumVersion: "14.0"
      - osType: DESKTOP_CHROME_OS
    requireScreenlock: true
    requireAdminApproval: true
    allowedDeviceManagementLevels:
      - ADVANCED
EOF

gcloud access-context-manager levels create managed-device \
  --policy=POLICY_ID \
  --title="Managed Device" \
  --basic-level-spec=managed-device.yaml

# Create access level for corporate network
cat > corp-network.yaml << 'EOF'
- ipSubnetworks:
    - "203.0.113.0/24"
    - "198.51.100.0/24"
  regions:
    - US
    - GB
EOF

gcloud access-context-manager levels create corp-network \
  --policy=POLICY_ID \
  --title="Corporate Network" \
  --basic-level-spec=corp-network.yaml

# Create custom access level using CEL for complex logic
cat > high-trust.yaml << 'EOF'
expression: >
  device.encryption_status == DeviceEncryptionStatus.ENCRYPTED &&
  device.is_admin_approved_device == true &&
  (
    origin.ip in ["203.0.113.0/24"] ||
    device.os_type == OsType.DESKTOP_CHROME_OS
  ) &&
  request.auth.claims.hd == "company.com"
EOF

gcloud access-context-manager levels create high-trust \
  --policy=POLICY_ID \
  --title="High Trust" \
  --custom-level-spec=high-trust.yaml

Step 4: Configure Session Settings and Re-authentication

Set session duration and re-authentication policies per application.

# Configure re-authentication for a backend service
# Requires login every 4 hours for sensitive apps
gcloud iap settings set \
  --project=PROJECT_ID \
  --resource-type=compute \
  --service=finance-app \
  reauthSettings.method=LOGIN \
  reauthSettings.maxAge=14400s \
  reauthSettings.policyType=MINIMUM

# Configure session settings for App Engine
gcloud iap settings set \
  --project=PROJECT_ID \
  --resource-type=app-engine \
  reauthSettings.method=SECURE_KEY \
  reauthSettings.maxAge=3600s \
  reauthSettings.policyType=MINIMUM

# View current IAP settings
gcloud iap settings get \
  --project=PROJECT_ID \
  --resource-type=compute \
  --service=finance-app

Step 5: Configure Programmatic Access for Service Accounts

Enable service-to-service communication through IAP-protected endpoints.

#!/usr/bin/env python3
"""Access IAP-protected resource using service account credentials."""

import google.auth
import google.auth.transport.requests
from google.auth import impersonated_credentials
import requests as req

IAP_CLIENT_ID = "YOUR_IAP_OAUTH_CLIENT_ID.apps.googleusercontent.com"
IAP_URL = "https://my-app.company.com/api/data"

def access_iap_resource():
    # Get default credentials (works with service account key or workload identity)
    credentials, project = google.auth.default()

    # Create IAP-authenticated request
    authed_session = google.auth.transport.requests.AuthorizedSession(
        credentials,
        target_audience=IAP_CLIENT_ID
    )

    # Make request to IAP-protected resource
    response = authed_session.get(IAP_URL)
    print(f"Status: {response.status_code}")
    print(f"Response: {response.text[:500]}")

    return response

if __name__ == "__main__":
    access_iap_resource()

Step 6: Set Up Audit Logging and Monitoring

Configure logging for all IAP access decisions.

# Enable data access audit logs for IAP
gcloud projects get-iam-policy PROJECT_ID --format=json > policy.json

# Add IAP audit config to policy.json:
# {
#   "service": "iap.googleapis.com",
#   "auditLogConfigs": [
#     {"logType": "ADMIN_READ"},
#     {"logType": "DATA_READ"},
#     {"logType": "DATA_WRITE"}
#   ]
# }

gcloud projects set-iam-policy PROJECT_ID policy.json

# Create log-based metric for denied access
gcloud logging metrics create iap-denied-access \
  --description="Count of IAP access denials" \
  --log-filter='resource.type="gce_backend_service" AND protoPayload.status.code=16'

# Create alerting policy for high denial rates
gcloud alpha monitoring policies create \
  --display-name="IAP High Denial Rate" \
  --condition-display-name="Denied access > 50 in 5 min" \
  --condition-filter='metric.type="logging.googleapis.com/user/iap-denied-access"' \
  --condition-threshold-value=50 \
  --condition-threshold-duration=300s \
  --notification-channels=projects/PROJECT_ID/notificationChannels/CHANNEL_ID

# Query IAP access logs
gcloud logging read '
  resource.type="gce_backend_service"
  protoPayload.serviceName="iap.googleapis.com"
  timestamp >= "2026-02-22T00:00:00Z"
' --project=PROJECT_ID --format='table(timestamp,protoPayload.authenticationInfo.principalEmail,protoPayload.status.code,resource.labels.backend_service_name)' --limit=50

Key Concepts

TermDefinition
Identity-Aware ProxyGCP service that intercepts web requests and TCP connections, authenticating users and evaluating access policies before proxying to backend services
Backend ServiceGCP load balancer component that IAP protects; can serve Compute Engine instances, GKE pods, Cloud Run services, or App Engine
IAP TunnelSecure TCP tunnel through IAP allowing SSH, RDP, and other TCP access to VMs without public IPs or VPN
OAuth Consent ScreenGCP configuration specifying the application name and support email shown to users during IAP authentication
Access LevelNamed condition in Access Context Manager evaluated during IAP authorization (device posture, IP, geography)
Re-authenticationIAP feature requiring users to prove their identity again after a configurable session duration

Tools & Systems

  • Google Cloud IAP: Identity-aware reverse proxy for GCP applications and TCP services
  • Access Context Manager: Defines access levels based on device, network, and geographic attributes
  • gcloud CLI: Command-line tool for configuring IAP, access levels, and IAM bindings
  • IAP TCP Forwarding: Tunnel-based access to VMs for SSH/RDP without public IPs
  • Cloud Audit Logs: Immutable records of all IAP access decisions for compliance
  • Endpoint Verification: Chrome extension collecting device attributes for access level evaluation

Common Scenarios

Scenario: Securing 15 Internal GCP Services with IAP

Context: An e-commerce company runs 15 internal services on GKE and Cloud Run (admin dashboards, internal APIs, monitoring tools). Currently, these services are protected only by VPN and firewall rules, creating excessive network-level access.

Approach:

  1. Deploy all services behind an HTTPS Load Balancer with managed SSL certificates
  2. Enable IAP on each backend service with per-service OAuth clients
  3. Create IAM bindings mapping Google Groups to specific services (admin group -> admin dashboard, engineering -> monitoring)
  4. Define access levels: managed-device (encryption + screen lock), corp-network (office IP ranges)
  5. Apply managed-device access level to admin dashboard and financial tools
  6. Configure IAP TCP tunneling for SSH access to GKE nodes (replacing SSH bastion host)
  7. Set re-authentication to 4 hours for admin tools, 8 hours for monitoring
  8. Configure Cloud Audit Logs and create alerting for repeated denials

Pitfalls: IAP adds 10-50ms latency per request; test application performance. WebSocket connections through IAP require specific backend service configuration. Service-to-service calls within GKE should bypass IAP using internal service mesh, not external IAP endpoints. Break-glass access should use a separate IAM binding without access level conditions.

Output Format

Google Cloud IAP Configuration Report
==================================================
Project: ecommerce-internal
Report Date: 2026-02-23

IAP-PROTECTED SERVICES:
  Backend Services:     12
  App Engine:            1
  Cloud Run:             2
  IAP TCP Tunnels:       4 (SSH access)
  Total:                19

ACCESS CONTROL:
  IAM Bindings:         34
  With Access Levels:   18 (52.9%)
  Access Levels:         3 (managed-device, corp-network, high-trust)

SESSION POLICIES:
  Admin tools:          4h re-auth (SECURE_KEY)
  Sensitive apps:       4h re-auth (LOGIN)
  General tools:        8h re-auth (LOGIN)

ACCESS LOGS (last 24h):
  Total requests:       23,456
  Authenticated:        23,289 (99.3%)
  Denied by IAM:           112
  Denied by access level:   55
  Unique users:            134

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Extract and decrypt Windows DPAPI-protected secrets (Credential Manager, browser logins/cookies, Wi-Fi credentials, KeePass keys) online or offline using SharpDPAPI, SharpChrome, Mimikatz, or Impacket's dpapi.py, including domain-wide decryption via the DPAPI backup key. Use during authorized red-team credential-access engagements after gaining a foothold or when triaging DPAPI blobs pulled from a host.

日本語の概要は準備中です。原文の説明を表示しています。

mukul975/Anthropic-Cybersecurity-Skills3.4万2026年8月31日 更新

Take over Active Directory accounts by writing attacker-controlled public keys to msDS-KeyCredentialLink (Shadow Credentials) with pyWhisker, Whisker, or Certipy, then authenticate via PKINIT to recover the target's NT hash without a password reset. Use when BloodHound shows GenericWrite/GenericAll/AddKeyCredentialLink over a target, as a stealthier alternative to ForceChangePassword, during authorized red-team engagements.

日本語の概要は準備中です。原文の説明を表示しています。

mukul975/Anthropic-Cybersecurity-Skills3.4万2026年8月31日 更新

Prepare a defense-contractor environment for CMMC Level 2 certification: scope CUI and FCI, implement the 110 NIST SP 800-171 Rev 2 security requirements across 14 families, compute the SPRS score with the DoD Assessment Methodology, manage a compliant POA&M, and ready the organization for a C3PAO assessment. Use when an organization handles Controlled Unclassified Information (CUI) under a DoD contract, when a contract carries DFARS clause 252.204-7012/7019/7020/7021, when preparing for or responding to a CMMC assessment, when computing or improving an SPRS score, when building a System Security Plan or POA&M for 800-171, or when scoping which systems are in the CUI boundary. Keywords: CMMC, CMMC Level 2, NIST 800-171, SP 800-171 Rev 2, CUI, FCI, SPRS, DFARS 7012, C3PAO, POA&M, System Security Plan, DoD Assessment Methodology, 110 controls, defense industrial base, DIB, FedRAMP equivalency.

日本語の概要は準備中です。原文の説明を表示しています。

mukul975/Anthropic-Cybersecurity-Skills3.4万2026年8月31日 更新

Create forensically sound bit-for-bit disk images with dd or dcfldd on a Linux forensic workstation, preserving evidence integrity through hash verification (MD5/SHA) during acquisition. Use when imaging a suspect drive, USB device, or memory card for investigation, preserving volatile disk evidence during incident response, or producing a verified copy for legal or law-enforcement proceedings before any destructive analysis.

日本語の概要は準備中です。原文の説明を表示しています。

mukul975/Anthropic-Cybersecurity-Skills3.4万2026年8月31日 更新

Detect dangerous ACL misconfigurations in Active Directory using ldap3 to identify GenericAll, WriteDACL, and WriteOwner abuse paths

日本語の概要は準備中です。原文の説明を表示しています。

mukul975/Anthropic-Cybersecurity-Skills3.4万2026年8月31日 更新

Perform static analysis of Android APK malware using apktool for resource decompilation, jadx for Java source recovery, and androguard for manifest inspection, dangerous permission-combination detection, and identification of obfuscated code, dynamic code loading, and reflection-based API calls. Use to statically triage a suspicious APK without executing it or to build mobile malware detection rules.

日本語の概要は準備中です。原文の説明を表示しています。

mukul975/Anthropic-Cybersecurity-Skills3.4万2026年8月31日 更新

mukul975 のスキルをすべて見る

このスキルの問題を報告する