安全扫描（ClawHub）

• VirusTotal ：良性
• OpenClaw ：良性

OpenClaw 评估

This is an instruction-only SOC 2 compliance guidance skill whose requested assets and runtime instructions match its stated purpose and do not ask for credentials or install code.

安装（复制给龙虾 AI）

将下方整段复制到龙虾中文库对话中，由龙虾按 SKILL.md 完成安装。

请把本段交给龙虾中文库（龙虾 AI）执行：为本机安装 OpenClaw 技能「SOC 2 AI Agent Compliance」。简介：Guides organizations through SOC 2 compliance lifecycle with gap analysis, cont…。
请 fetch 以下地址读取 SKILL.md 并按文档完成安装：https://raw.githubusercontent.com/openclaw/skills/refs/heads/main/skills/1kalin/afrexai-soc2-compliance/SKILL.md
（来源：yingzhi8.cn 技能库）

SKILL.md

打开原始 SKILL.md（GitHub raw）

# SOC 2 Compliance Accelerator

Your agent for achieving and maintaining SOC 2 Type I and Type II compliance — from readiness assessment through audit completion.

## What This Does

Guides organizations through the full SOC 2 lifecycle: gap analysis, control implementation, evidence collection, audit prep, and continuous monitoring. Covers all 5 Trust Service Criteria with practical implementation steps.

## How to Use

Tell your agent what stage you're at:

- **"Run SOC 2 readiness assessment"** — 64-point gap analysis across all Trust Service Criteria
- **"Build SOC 2 control matrix"** — Maps controls to criteria with ownership and evidence requirements
- **"Create SOC 2 evidence collection plan"** — Automated and manual evidence gathering schedule
- **"Prepare for SOC 2 audit"** — Auditor-ready documentation package checklist
- **"SOC 2 continuous monitoring dashboard"** — Ongoing compliance tracking after certification

## Trust Service Criteria Coverage

### CC — Common Criteria (Security) — Required
- CC1: Control Environment (tone at top, org structure, accountability)
- CC2: Communication & Information (internal/external, system boundaries)
- CC3: Risk Assessment (risk identification, fraud risk, change impact)
- CC4: Monitoring Activities (ongoing evaluations, deficiency reporting)
- CC5: Control Activities (policies, technology controls, deployment)
- CC6: Logical & Physical Access (access management, authentication, physical security)
- CC7: System Operations (vulnerability management, incident response, recovery)
- CC8: Change Management (change authorization, testing, approval)
- CC9: Risk Mitigation (vendor management, business continuity)

### Optional Criteria
- **Availability (A1)**: Uptime SLAs, DR/BCP, capacity planning
- **Processing Integrity (PI1)**: Data accuracy, completeness, timeliness
- **Confidentiality (C1)**: Classification, encryption, retention, disposal
- **Privacy (P1)**: Notice, consent, collection, use, disclosure, access

## Readiness Assessment Framework

### Phase 1: Scoping (Week 1)
```
System Description Checklist:
□ Infrastructure components (cloud, on-prem, hybrid)
□ Software stack (applications, databases, middleware)
□ People (roles, responsibilities, third parties)
□ Procedures (operational, security, change management)
□ Data flows (ingress, processing, storage, egress)
□ Trust Service Criteria selection (Security + which optional?)
□ Subservice organizations (cloud providers, SaaS tools)
□ Carve-out vs inclusive method for subservice orgs
```

### Phase 2: Gap Analysis (Weeks 2-3)
Score each control area 1-5:
- **1 — Not Started**: No policy, no process, no evidence
- **2 — Ad Hoc**: Informal processes exist but undocumented
- **3 — Defined**: Documented but inconsistent execution
- **4 — Managed**: Documented, executed, some evidence
- **5 — Optimized**: Automated, monitored, auditable evidence

Priority Matrix:
| Gap Score | Action | Timeline |
|-----------|--------|----------|
| 1-2 | Critical — implement immediately | 2-4 weeks |
| 3 | Important — formalize and document | 1-2 weeks |
| 4 | Minor — fill evidence gaps | 3-5 days |
| 5 | Maintain — continue monitoring | Ongoing |

### Phase 3: Remediation (Weeks 3-10)
```
For each gap:
1. Assign control owner (by name, not role)
2. Define implementation steps
3. Set evidence collection method (automated preferred)
4. Establish testing cadence
5. Document exception handling process
```

## Control Implementation Priorities

### Must-Have Controls (Week 1-4)
1. **Access Management**: SSO, MFA on all systems, quarterly access reviews
2. **Encryption**: TLS 1.2+ in transit, AES-256 at rest, key management
3. **Logging**: Centralized logging, 90-day retention minimum, tamper-evident
4. **Incident Response**: Documented plan, defined roles, tested annually
5. **Change Management**: Approval workflows, code review, deployment gates
6. **Vendor Management**: Vendor inventory, risk assessments, SOC 2 reports from critical vendors
7. **Employee Security**: Background checks, security awareness training, acceptable use policy
8. **Vulnerability Management**: Regular scanning, patch cadence (critical <72hrs), penetration testing

### Should-Have Controls (Week 4-8)
9. **Business Continuity**: DR plan, RTO/RPO defined, tested semi-annually
10. **Data Classification**: 4-tier model (Public, Internal, Confidential, Restricted)
11. **Network Security**: Segmentation, IDS/IPS, WAF for web applications
12. **Endpoint Protection**: EDR, device encryption, MDM for mobile

### Nice-to-Have Controls (Week 8+)
13. **Security Metrics Dashboard**: Real-time compliance posture
14. **Automated Compliance Monitoring**: Continuous control testing
15. **Zero Trust Architecture**: Beyond perimeter security

## Evidence Collection Guide

### Automated Evidence (Set Once, Collect Forever)
| Control | Evidence Source | Tool Examples |
|---------|---------------|---------------|
| Access Reviews | IAM exports | Okta, Azure AD, AWS IAM |
| Encryption | Config snapshots | AWS Config, CloudTrail |
| Logging | Log aggregation | Datadog, Splunk, ELK |
| Vulnerability Scans | Scan reports | Qualys, Nessus, Snyk |
| Change Management | PR/deploy history | GitHub, GitLab, Jira |
| Uptime | Monitoring dashboards | Datadog, PagerDuty |

### Manual Evidence (Scheduled Collection)
| Control | Evidence Type | Frequency |
|---------|--------------|-----------|
| Background Checks | HR records | Per hire |
| Security Training | Completion certificates | Annual |
| Risk Assessment | Assessment document | Annual |
| Pen Testing | Report | Annual |
| DR Testing | Test results | Semi-annual |
| Board/Mgmt Review | Meeting minutes | Quarterly |
| Vendor Reviews | Assessment records | Annual |
| Policy Reviews | Version history | Annual |

## Audit Timeline

### Type I (Point-in-Time) — 8-12 weeks total
```
Week 1-2:   Auditor selection + engagement letter
Week 2-4:   System description draft
Week 4-6:   Control documentation + evidence prep
Week 6-8:   Fieldwork (auditor testing)
Week 8-10:  Draft report review
Week 10-12: Final report issued
```

### Type II (Period of Time) — 3-12 month observation + 4-6 weeks fieldwork
```
Month 1:     Observation period begins (minimum 3 months, recommend 6-12)
Ongoing:     Evidence collection, control operation
Month 3-12:  Observation period ends
+Week 1-2:   Fieldwork scheduling
+Week 2-4:   Fieldwork (testing over observation period)
+Week 4-6:   Draft report + final report
```

## Cost Framework

| Company Size | Type I | Type II | Annual Maintenance |
|-------------|--------|---------|-------------------|
| Startup (<50) | $20K-$50K | $30K-$80K | $15K-$40K |
| Mid-Market (50-500) | $40K-$100K | $60K-$150K | $30K-$80K |
| Enterprise (500+) | $80K-$200K | $120K-$300K | $60K-$150K |

Includes: auditor fees, tooling, personnel time, remediation costs.

Hidden costs to budget:
- Compliance automation platform: $10K-$50K/year
- Additional security tooling: $5K-$30K/year
- Personnel time (internal): 200-800 hours
- Policy/procedure writing (if outsourced): $5K-$20K

## Common Audit Findings (Avoid These)

1. **Access not revoked within 24 hours of termination** — #1 finding
2. **Missing or incomplete risk assessment** — annual requirement
3. **No evidence of management review** — need meeting minutes
4. **Incomplete vendor management** — missing SOC reports from critical vendors
5. **Inconsistent change management** — emergency changes without retroactive approval
6. **Security training gaps** — new hires not trained within 30 days
7. **Logging gaps** — not all in-scope systems sending to central logging

## AI Agent SOC 2 Considerations (2026)

When deploying AI agents in SOC 2 environments:
- **Data boundaries**: Agents must not access data outside their defined scope
- **Audit trail**: All agent actions must be logged and attributable
- **Access controls**: Agent service accounts need same rigor as human accounts
- **Model governance**: Document which models process customer data
- **Prompt injection defense**: Part of CC7 (system operations) controls
- **Output validation**: Processing integrity controls for agent outputs

## Industry-Specific Requirements

| Industry | Extra Criteria | Key Controls |
|----------|---------------|-------------|
| **Fintech** | All 5 TSC typical | SOX mapping, encryption everywhere, PCI if payments |
| **Healthcare** | Privacy, Confidentiality | HIPAA crosswalk, BAAs, PHI handling |
| **SaaS** | Availability, Confidentiality | Multi-tenant isolation, SLA compliance |
| **Legal** | Confidentiality, Privacy | Privilege protection, matter isolation |
| **Construction** | Security, Availability | Field data protection, offline capability |
| **E-commerce** | All 5 TSC typical | PCI DSS alignment, transaction integrity |

## 7 SOC 2 Mistakes That Cost Companies 6+ Months

1. **Starting with Type II** — Get Type I first, prove controls work, then observe
2. **Scoping too broadly** — Only include systems that touch customer data
3. **Choosing the wrong auditor** — Pick one who knows your industry
4. **Manual evidence collection** — Automate from day 1 or drown in spreadsheets
5. **Treating it as a project, not a program** — SOC 2 is continuous
6. **Ignoring subservice organizations** — Your cloud provider's SOC 2 matters
7. **No executive sponsor** — Compliance without budget authority = failure

---

## Get the Full Implementation Package

This skill gives you the framework. For industry-specific compliance playbooks with regulatory crosswalks, cost models, and vendor selection guides:

🔗 **[AfrexAI Context Packs](https://afrexai-cto.github.io/context-packs/)** — $47 per industry vertical

Available packs: Fintech, Healthcare, Legal, Construction, E-commerce, SaaS, Real Estate, Recruitment, Manufacturing, Professional Services

🔗 **[AI Revenue Leak Calculator](https://afrexai-cto.github.io/ai-revenue-calculator/)** — Find where compliance gaps cost you money

🔗 **[Agent Setup Wizard](https://afrexai-cto.github.io/agent-setup/)** — Deploy compliance monitoring agents in minutes

**Bundle pricing:**
- Pick 3 packs: $97
- All 10 packs: $197
- Everything bundle: $247