CCC Managed Vector Store Controls
Controls for Managed Vector Store technologies, as defined by the FINOS Common Cloud Controls project.
- ID
- CCC.Vector.CN
- Version
- v2026.06-rc3
- Gemara version
- v1.2.0
- Author
- FINOS Common Cloud Controls
Ingestion
The Ingestion group covers entries related to how a service receives or retrieves data, inputs, or commands for processing. This includes both active (pull-based) and passive (push-based) ingestion patterns.
CCC.Vector.CN01 Validate Embeddings Before Indexing
Objective
Ensure all incoming embeddings are structurally and statistically validated before indexing to prevent poisoning or corruption.
Assessment requirements
When a vector embedding is submitted for indexing, the system MUST validate that it matches expected schema, dimension, and format profiles.
Applicability: tlp-clear, tlp-green, tlp-amber, tlp-red
Guidelines
- FINOS-AIGF
- AIR-PREV-002 — Data Filtering From External Knowledge Bases
Threats
- CCC.Vector.Threats
- CCC.Vector.TH02 — Embedding and Index Poisoning
- CCC.Vector.TH05 — Embedding Format or Dimension Attacks
- CCC.Core.Threats
- CCC.Core.TH12 — Resource Constraints are Exhausted
CCC.Vector.CN04 Enforce Ingestion Quotas and Throttling
Objective
Prevent ingestion-based DoS or index pollution by rate-limiting vector submissions and enforcing quotas.
Assessment requirements
When ingestion exceeds pre-defined thresholds, the service MUST throttle or reject excess vector write operations.
Applicability: tlp-green, tlp-amber, tlp-red
Guidelines
- FINOS-AIGF
- AIR-PREV-008 — Quality of Service (QoS) and DDoS Prevention for AI Systems
Threats
- CCC.Vector.Threats
- CCC.Vector.TH02 — Embedding and Index Poisoning
- CCC.Core.Threats
- CCC.Core.TH12 — Resource Constraints are Exhausted
CCC.Vector.CN06 Enforce Dimensional and Format Constraints
Objective
Reject embeddings that do not conform to expected model specifications (dimensions, format, etc).
Assessment requirements
When an embedding is submitted, the service MUST validate that its format and dimensionality match allowed profiles.
Applicability: tlp-clear, tlp-green, tlp-amber, tlp-red
Guidelines
- FINOS-AIGF
- AIR-PREV-002 — Data Filtering From External Knowledge Bases
Threats
- CCC.Vector.Threats
- CCC.Vector.TH05 — Embedding Format or Dimension Attacks
- CCC.Core.Threats
- CCC.Core.TH06 — Data is Lost or Corrupted
Access Control
The Access Control group covers entries related to authentication, authorization, and trust perimeter enforcement. This includes multi-factor authentication, least privilege access, network access rules, and prevention of unauthorized access or reconnaissance.
CCC.Vector.CN02 Enforce Role-Based Index Lifecycle Management
Objective
Restrict index lifecycle operations (create, delete, rollback) to privileged identities using fine-grained access controls.
Assessment requirements
When an index lifecycle event is triggered, the service MUST verify that the actor has explicit permissions for the operation type.
Applicability: tlp-clear, tlp-green, tlp-amber, tlp-red
Guidelines
- FINOS-AIGF
- AIR-PREV-012 — Role-Based Access Control for AI Data
Threats
- CCC.Vector.Threats
- CCC.Vector.TH02 — Embedding and Index Poisoning
- CCC.Vector.TH04 — Index Corruption or Downgrade
- CCC.Core.Threats
- CCC.Core.TH01 — Access Control is Misconfigured
CCC.Vector.CN03 Enforce Metadata-Level Access Controls
Objective
Apply access control policies to metadata fields used in filtering to prevent unauthorized exposure or inference.
Assessment requirements
When a metadata filter is applied to a query, the service MUST verify the requester is authorized to access that field.
Applicability: tlp-amber, tlp-red
Guidelines
- FINOS-AIGF
- AIR-DET-001 — AI Data Leakage Prevention and Detection
- AIR-PREV-012 — Role-Based Access Control for AI Data
- AIR-DET-016 — Preserving Source Data Access Controls in AI Systems
Threats
- CCC.Vector.Threats
- CCC.Vector.TH03 — Cross-modal or Metadata Leakage
- CCC.Core.Threats
- CCC.Core.TH01 — Access Control is Misconfigured
Resource Management
The Resource Management group covers entries related to the lifecycle, configuration, and operational integrity of cloud resources. This includes resource exhaustion, tag manipulation, version rollback, scaling, and cost management.
CCC.Vector.CN05 Enforce Index Versioning with Rollback Protection
Objective
Ensure vector indexes are versioned and that rollback operations are authorized and auditable.
Assessment requirements
When a rollback is attempted, the system MUST log the action and verify rollback authorization.
Applicability: tlp-amber, tlp-red
Guidelines
- AIR-DET-004
- AIR-PREV-008 — AI System Observability
Threats
- CCC.Vector.Threats
- CCC.Vector.TH04 — Index Corruption or Downgrade
- CCC.Core.Threats
- CCC.Core.TH09 — Logs or Monitoring Data are Read by Unauthorized Users
- CCC.Core.TH04 — Data is Replicated to Untrusted or External Locations
Data Processing
The Data Processing group covers entries related to transforming, enriching, and moving data through pipelines. This includes ETL/ELT, stream and batch processing, data lineage, schema evolution, and workflow orchestration for data workloads.
CCC.Vector.CN07 Support Explicit ANN vs. Exact Search Configuration
Objective
Provide clients with the option to enforce exact-match (non-ANN) search where search fidelity is critical.
Assessment requirements
When a search request is issued, clients MUST be allowed to declare their requirement for exact vs approximate results.
Applicability: tlp-amber, tlp-red
Threats
- CCC.Vector.Threats
- CCC.Vector.TH06 — Search Result Manipulation via ANN Bias