Use cases and risk areas to investigate¶

1. Areas to explore for privacy risks¶

• Personal/sensitive data storage
• Egress to openclaw orchestrator and models (perplexity, anthropic, others) (Prio 1)
• Export of reports to storage
• Role base access controls (Prio 2)
• Telemetry: evaluation user workflow tracking (Prio 3)
• Structured logging: coding workflow tracking (Prio 3)
• Secrets, tokens, and keys (Prio 4)
• CI

2. General application workflow¶

• User interacts with application
• User logs in
• User uploads document
• Document is persisted in DB
• Document is analyzed by AI
• AI suggests risk templates
• AI suggests document analysis
• AI suggests human review
• Human review process HITL (Right now always the same user, because it's connected to the login. May change in the future.)
• Store reviewed document
• Store reports
• Export reports to recipients

3. General software processes and services¶

• Access controls (RBAC)
• Authentication
• Authorization
• Logging
• Telemetry
• Secrets, keys, and tokens
• CI/CD
• Docker containerization

4. Data storage¶

Database tables for application (ORM: src/doc_quality/models/orm.py)¶

• app_users
• user_sessions
• password_recovery_tokens
• skill_documents
• hitl_reviews
• bridge_human_reviews
• audit_events
• audit_schedules
• skill_findings
• quality_observations
• stakeholder_employee_assignments
• risk_templates
• document_locks

Auth-related audit payloads (src/doc_quality/api/routes/auth.py)¶

• Logins
• password recovery
• cookies
• session tokens

Structured logging¶

• Application logs
  • errors
  • requests
  • status events
• Spans for tracing
• Document skills output persisted
• Access logs
• AI services: document excerpts, errors, events, and logs

5. AI orchestrator and models — Anthropic and Perplexity¶

OpenClaw¶

• Orchestrate messages to AI agents

Perplexity¶

• Requests
• Prompts
• Responses
• Metadata
• Payloads

Anthropic¶

• Requests
• Prompts
• Responses
• Metadata
• Payloads

Agents: - Document analysis agent (doc_check_agent) - Compliance agent (compliance_agent)

Risk mitigation: - in example template ->>> to be removed

6. Exports, remote uploads, records retention¶

Export registry¶

• local downloads
• remote exports

Records retention¶

Who has access to what? How long?

• local
• offsite

Risk mitigation: - RBAC - admin account can define roles and access - storage is forever right now - admin can delete users and records

IDE / MCP / Copilot¶

What is used

• VS Code & GitHub Copilot

7. Access controls (individuals / org)¶

• RBAC
• Authenticated users and HITL reviewers
• Audit logs

Document and Human in the Loop (HITL)¶

• Document access
• HITL review access and actions

Secrets, tokens, and keys (src/doc_quality/core/config.py)¶

• Secret key for application
• DB access
• Anthropic and Perplexity API keys
• Auth logins
• Tokens
• Cookies

8. Logging, configuration, telemetry, CI, Docker¶

Logging¶

Structured logging and information collected.

Telemetry and metrics¶

• Traces and spans stored
• Data collection by Prometheus
• Metrics exposed to endpoints

Docker¶

Container logs and info to stdoout or stderr

CI¶

• action artifacts
• testing artifacts, like Playwright screenshots

9. Documentation and dev data¶

• Snippets of real data
• Mocked data for testing