Skip to content

MeshFlow SOC 2 Controls Mapping

This document maps MeshFlow's built-in controls to the SOC 2 Trust Services Criteria (TSC) under the AICPA's 2017 Trust Services Criteria framework.

CC1 — Control Environment

Criterion MeshFlow Control Evidence Location
CC1.1 — COSO Principle 1: Integrity and ethical values DID-based agent identity; collusion detection meshflow/security/identity.py, meshflow/intelligence/collusion.py
CC1.2 — Board oversight Policy mode enforcement; HITL gates meshflow/core/policy.py
CC1.3 — Organisational structure Agent role assignment; team patterns meshflow/agents/team.py
CC1.4 — Competence commitment Schema-validated tool calls; type enforcement meshflow/agents/base.py
CC1.5 — Accountability Immutable SHA-256 audit chain; chain validation meshflow/core/ledger.py

CC2 — Communication and Information

Criterion MeshFlow Control Evidence Location
CC2.1 — Information relevant to financial reporting Tamper-evident ledger; all steps recorded meshflow/core/ledger.py
CC2.2 — Internal communication /traces/{run_id} audit endpoint; CLI trace command meshflow/runtime/server.py, meshflow/cli/main.py
CC2.3 — External communication Webhook HITL notifications with HMAC-SHA256 meshflow/core/hitl.py

CC3 — Risk Assessment

Criterion MeshFlow Control Evidence Location
CC3.1 — Risk identification Uncertainty scoring; risk tier classification meshflow/core/schemas.py (RiskTier)
CC3.2 — Risk analysis DascGate policy enforcement; budget/token caps meshflow/security/dasc_gate.py
CC3.3 — Risk mitigation Guardian safety layer; collusion detection meshflow/security/guardian.py, meshflow/intelligence/collusion.py
CC3.4 — Change management Schema migration registry with version tracking meshflow/core/ledger.py (_MIGRATIONS)

CC4 — Monitoring Activities

Criterion MeshFlow Control Evidence Location
CC4.1 — Monitoring controls Prometheus metrics endpoint /metrics meshflow/observability/metrics.py
CC4.2 — Evaluating deficiencies Collusion alerts; uncertainty threshold alerts meshflow/intelligence/collusion.py

CC5 — Control Activities

Criterion MeshFlow Control Evidence Location
CC5.1 — Control selection Policy presets (dev/standard/regulated/legal-critical/hipaa) meshflow/core/schemas.py
CC5.2 — Technology controls API key authentication; Bearer token validation meshflow/runtime/server.py
CC5.3 — Deployment controls Docker multi-stage build; K8s health probes Dockerfile, k8s/deployment.yaml

CC6 — Logical and Physical Access

Criterion MeshFlow Control Evidence Location
CC6.1 — Access registration MESHFLOW_API_KEYS env var; key rotation support meshflow/runtime/server.py
CC6.2 — Authentication Bearer token or X-API-Key header auth meshflow/runtime/server.py (_check_auth)
CC6.3 — Access removal delete_tenant() removes all tenant data meshflow/core/ledger.py
CC6.6 — Logical access boundaries Multi-tenancy; per-tenant ledger scoping meshflow/core/ledger.py (tenant_id)
CC6.7 — Data transmission TLS support (--tls-cert, --tls-key flags) meshflow/runtime/server.py
CC6.8 — Malicious software PHI scrubbing; input sanitisation; shell blocklist meshflow/security/phi_scrubber.py, meshflow/tools/builtins.py

CC7 — System Operations

Criterion MeshFlow Control Evidence Location
CC7.1 — Configuration management Environment variable based config; policy presets meshflow/core/schemas.py
CC7.2 — Vulnerability management Tool call schema validation; safe eval (no eval()) meshflow/agents/base.py, meshflow/tools/builtins.py
CC7.3 — Incident identification Collusion detection; uncertainty alerts; HITL gates meshflow/intelligence/collusion.py
CC7.4 — Incident response HITL approve/reject; anonymize_run() meshflow/core/ledger.py, meshflow/core/hitl.py
CC7.5 — Disaster recovery Pluggable ledger backends (SQLite, PostgreSQL) meshflow/core/ledger.py

CC8 — Change Management

Criterion MeshFlow Control Evidence Location
CC8.1 — Change authorisation Schema migration registry; versioned migrations meshflow/core/ledger.py (_MIGRATIONS)

CC9 — Risk Mitigation

Criterion MeshFlow Control Evidence Location
CC9.1 — Risk mitigation activities Budget caps; max step limits; timeout enforcement meshflow/core/policy.py
CC9.2 — Business disruption risk K8s HPA (2–10 replicas); health probes k8s/deployment.yaml

Availability (A1)

Criterion MeshFlow Control Evidence Location
A1.1 — Availability commitments Health endpoint /health; liveness + readiness probes meshflow/runtime/server.py, k8s/deployment.yaml
A1.2 — Environmental safeguards Docker restart policy; K8s deployment with PVC docker-compose.yml, k8s/deployment.yaml
A1.3 — Recovery procedures Durable ledger (file/postgres); replay on restart meshflow/core/ledger.py

Confidentiality (C1)

Criterion MeshFlow Control Evidence Location
C1.1 — Confidential information PHI scrubber; output compression; tenant isolation meshflow/security/phi_scrubber.py
C1.2 — Confidential information disposal delete_run(), delete_tenant(), anonymize_run() meshflow/core/ledger.py

Privacy (P Series)

Criterion MeshFlow Control Evidence Location
P4.1 — Data minimisation max_output_chars policy field meshflow/core/schemas.py
P4.2 — Retention delete_run() for erasure requests meshflow/core/ledger.py
P6.1 — Data subject requests GDPR guide; anonymize_run() docs/compliance/GDPR_GUIDE.md
P8.1 — PHI handling scrub_phi policy flag; PHIScrubber meshflow/security/phi_scrubber.py

Audit Evidence Collection

To support a SOC 2 audit, collect the following artefacts:

  1. Ledger export — meshflow trace <run-id> --export for sampled runs.
  2. Chain validation — meshflow trace <run-id> showing CHAIN VALID.
  3. Prometheus metrics snapshot — GET /metrics at audit date.
  4. Policy configuration — export the Policy objects used in production.
  5. HITL decision log — export HITL approve/reject decisions from the ledger.
  6. Dependency audit — pip list --format=freeze from the production image.
  7. Container image digest — docker inspect meshflow:latest --format '{{.Id}}'.