Agent Skills: Performing SOC 2 Type II Audit Preparation

Performing SOC 2 Type II Audit Preparation

When to Use

• When preparing for a SOC 2 Type II audit engagement with a CPA firm
• When conducting a gap assessment against AICPA Trust Services Criteria
• When automating evidence collection across cloud infrastructure and identity providers
• When validating that controls have operated effectively over the audit period (3-12 months)
• When building continuous compliance monitoring to maintain SOC 2 posture between audits
• When remediating control gaps identified during readiness assessment

Prerequisites

• Familiarity with AICPA Trust Services Criteria (CC1-CC9)
• Access to cloud provider APIs (AWS, Azure, or GCP) with read-only permissions
• Access to identity provider (Okta, Azure AD, Google Workspace)
• Access to version control system (GitHub, GitLab)
• Access to ticketing system (Jira, Linear, ServiceNow)
• Python 3.8+ with boto3, requests, pyyaml dependencies
• Appropriate authorization to collect compliance evidence

Instructions

1. Understand the Trust Services Criteria

SOC 2 is built on five Trust Services Categories defined by AICPA. Security (Common Criteria CC1-CC9) is mandatory; the others are selected based on business relevance:

| Category | Criteria | Focus | |----------|----------|-------| | Security (mandatory) | CC1-CC9 | Control environment, risk, access, operations, change management | | Availability | A1 | System uptime and disaster recovery | | Processing Integrity | PI1 | Accurate and complete data processing | | Confidentiality | C1 | Protection of confidential information | | Privacy | P1-P8 | Personal information lifecycle |

2. Common Criteria Breakdown (CC1-CC9)

CC1 - Control Environment: Board oversight, management structure, integrity and ethical values, HR policies, accountability.

CC2 - Communication and Information: Internal/external communication of security policies, system boundaries, roles, and responsibilities.

CC3 - Risk Assessment: Risk identification, fraud risk analysis, change impact assessment, risk tolerance definition.

CC4 - Monitoring Activities: Ongoing control evaluations, deficiency identification, remediation tracking, internal audit.

CC5 - Control Activities: Policy-to-procedure mapping, technology controls, deployment of controls across the entity.

CC6 - Logical and Physical Access Controls: Authentication, authorization, access provisioning/deprovisioning, physical security, encryption.

CC7 - System Operations: Anomaly detection, incident response, vulnerability management, change detection, event monitoring.

CC8 - Change Management: Change authorization, testing, approval workflows, emergency changes, rollback procedures.

CC9 - Risk Mitigation: Vendor risk management, business continuity, insurance, residual risk acceptance.

3. Conduct Gap Assessment

Before the audit period begins, perform a readiness assessment 8-12 weeks in advance:

# Define control matrix against CC criteria
gap_assessment = {
    "CC1": {
        "CC1.1": {
            "criteria": "COSO Principle 1: Demonstrates commitment to integrity",
            "control": "Code of conduct signed annually by all employees",
            "evidence": "Signed acknowledgments in HR system",
            "status": "implemented",
            "gap": None,
        },
        "CC1.2": {
            "criteria": "COSO Principle 2: Board exercises oversight",
            "control": "Quarterly board security reviews",
            "evidence": "Board meeting minutes with security agenda items",
            "status": "partial",
            "gap": "No documented security committee charter",
        },
    },
}

4. Automate Evidence Collection

Collect evidence continuously throughout the audit period from integrated systems:

import boto3

# CC6 Evidence: AWS IAM access controls
iam = boto3.client("iam")

# Collect MFA status for all IAM users
users = iam.list_users()["Users"]
mfa_evidence = []
for user in users:
    mfa_devices = iam.list_mfa_devices(UserName=user["UserName"])
    mfa_evidence.append({
        "user": user["UserName"],
        "mfa_enabled": len(mfa_devices["MFADevices"]) > 0,
        "created": user["CreateDate"].isoformat(),
    })

# CC7 Evidence: AWS CloudTrail logging status
cloudtrail = boto3.client("cloudtrail")
trails = cloudtrail.describe_trails()["trailList"]
logging_evidence = []
for trail in trails:
    status = cloudtrail.get_trail_status(Name=trail["TrailARN"])
    logging_evidence.append({
        "trail": trail["Name"],
        "is_logging": status["IsLogging"],
        "multi_region": trail.get("IsMultiRegionTrail", False),
        "log_validation": trail.get("LogFileValidationEnabled", False),
    })

5. Validate Control Effectiveness

For Type II audits, demonstrate controls operated effectively over the entire audit period:

import requests

# CC8 Evidence: Change management - verify all production changes
# had tickets, approvals, and testing before deployment
headers = {"Authorization": f"token {github_token}"}
prs = requests.get(
    "https://api.github.com/repos/org/repo/pulls",
    params={"state": "closed", "base": "main", "per_page": 100},
    headers=headers,
).json()

change_evidence = []
for pr in prs:
    if not pr.get("merged_at"):
        continue
    reviews = requests.get(pr["url"] + "/reviews", headers=headers).json()
    approved = any(r["state"] == "APPROVED" for r in reviews)
    change_evidence.append({
        "pr_number": pr["number"],
        "title": pr["title"],
        "merged_at": pr["merged_at"],
        "approved": approved,
    })

# Flag PRs merged without approval (control exception)
exceptions = [c for c in change_evidence if not c["approved"]]

6. Continuous Compliance Monitoring

Set up automated checks that run daily to detect control drift:

# Daily compliance check - run via cron or Lambda
checks = [
    {"control": "CC6.1", "check": "All IAM users have MFA enabled"},
    {"control": "CC6.6", "check": "No public S3 buckets"},
    {"control": "CC7.1", "check": "CloudTrail logging enabled"},
    {"control": "CC7.2", "check": "GuardDuty findings under threshold"},
    {"control": "CC8.1", "check": "All PRs have required reviews"},
]

for check in checks:
    result = run_compliance_check(check["control"])
    if not result["passing"]:
        send_alert(
            channel="#compliance",
            message=f"Control drift: {check['control']} - {check['check']}",
            details=result["findings"],
        )

7. Prepare Evidence Packages for Auditors

Organize collected evidence into structured packages per criteria:

evidence_package = {
    "audit_period": {"start": "2025-04-01", "end": "2026-03-31"},
    "criteria_packages": {
        "CC1_Control_Environment": {
            "CC1.1": ["signed_acknowledgments.csv"],
            "CC1.2": ["board_minutes_q1.pdf", "board_minutes_q2.pdf"],
        },
        "CC6_Logical_Physical_Access": {
            "CC6.1": ["okta_mfa_policy.json", "iam_users_mfa_status.csv"],
            "CC6.2": ["access_review_q1.csv", "access_review_q2.csv"],
            "CC6.3": ["offboarding_tickets.csv", "terminated_user_audit.csv"],
        },
        "CC7_System_Operations": {
            "CC7.1": ["cloudtrail_config.json", "siem_dashboard.png"],
            "CC7.2": ["guardduty_findings_summary.csv"],
            "CC7.3": ["vulnerability_scan_reports/"],
        },
        "CC8_Change_Management": {
            "CC8.1": ["merged_prs_with_approvals.csv"],
        },
    },
}

Examples

Automated Access Review for CC6.2

import boto3
from datetime import datetime, timedelta

iam = boto3.client("iam")

# Find users with no activity in 90 days
inactive_threshold = datetime.utcnow() - timedelta(days=90)
report = iam.get_credential_report()["Content"].decode()

inactive_users = []
for line in report.strip().split("\n")[1:]:
    fields = line.split(",")
    username = fields[0]
    last_used = fields[4]
    if last_used not in ("N/A", "no_information"):
        last_date = datetime.strptime(last_used, "%Y-%m-%dT%H:%M:%S+00:00")
        if last_date < inactive_threshold:
            inactive_users.append({"user": username, "last_active": last_used})

Vulnerability Management Evidence for CC7.2

import requests

headers = {"Authorization": f"Bearer {scanner_token}"}
scans = requests.get(
    "https://scanner.example.com/api/v1/scans",
    params={"status": "completed", "since": "2025-04-01"},
    headers=headers,
).json()

vuln_evidence = {"scan_count": len(scans), "critical_findings": 0, "high_findings": 0}
for scan in scans:
    findings = requests.get(
        f"https://scanner.example.com/api/v1/scans/{scan['id']}/findings",
        headers=headers,
    ).json()
    vuln_evidence["critical_findings"] += len([f for f in findings if f["severity"] == "critical"])
    vuln_evidence["high_findings"] += len([f for f in findings if f["severity"] == "high"])

Incident Response Evidence for CC7.3

incidents = requests.get(
    "https://pagerduty.com/api/v1/incidents",
    params={"since": "2025-04-01", "until": "2026-03-31"},
    headers={"Authorization": f"Token token={pd_token}"},
).json()

ir_evidence = {
    "total_incidents": len(incidents["incidents"]),
    "incidents": [
        {
            "id": inc["id"],
            "title": inc["title"],
            "severity": inc["urgency"],
            "created": inc["created_at"],
            "resolved": inc.get("last_status_change_at"),
        }
        for inc in incidents["incidents"]
    ],
}