VB-OS Connectors: Evidence Acquisition Overview
Connectors bridge external systems and VB-OS boundaries. They acquire evidence from data sources, transform it into the format boundaries expect, and attach provenance metadata that records where each piece of evidence came from.
How Connectors Work
Section titled “How Connectors Work”- Connect: configure a connector with credentials for an external system
- Map: define how source fields map to boundary evidence fields
- Gather: the connector queries the source and produces evidence
- Evaluate: the gathered evidence is submitted to a boundary for evaluation
Acquisition Classes
Section titled “Acquisition Classes”Every piece of evidence has an acquisition class that describes how it entered VB-OS:
| Class | Description |
|---|---|
active_provider |
The connector actively queried the external system |
authenticated_provider_push |
The external system pushed data via a verified webhook |
caller_supplied_payload |
The caller submitted evidence directly in the API request |
Acquisition class is part of provenance: boundaries can require specific acquisition classes through provenance declarations.
Provider Categories
Section titled “Provider Categories”VB-OS includes a registry of providers across domains:
- Databases: PostgreSQL, MySQL, and other SQL databases
- Payment: Stripe, payment processors
- Identity: identity verification providers
- AI/ML: AI inference and model deployment providers
- DevOps: CI/CD, monitoring, deployment tools
- Healthcare: FHIR-compatible health systems
- Compliance: regulatory and compliance data sources
- And more: security, analytics, CRM, ERP, cloud infrastructure
Each provider defines its authentication modes, evidence fields, and capabilities.
Authentication Modes
Section titled “Authentication Modes”Connectors support multiple authentication patterns:
| Mode | Description |
|---|---|
CREDENTIAL |
API keys, connection strings, or tokens stored encrypted |
OAUTH_USER_CONSENT |
OAuth 2.0 authorization code flow with platform-managed tokens |
NONE |
No authentication required (public endpoints) |
Credentials are encrypted at rest. OAuth tokens are managed by the platform with automatic refresh.
Evidence Mapping
Section titled “Evidence Mapping”Evidence mappings define how connector source data maps to boundary evidence fields:
evidence_field: the target field name in the boundarysource_path: dot-notation path into the provider’s response (e.g.,data.status)transform: optional transformation applied during mapping
Transforms
Section titled “Transforms”| Transform | Description |
|---|---|
identity |
Pass through unchanged |
to_string |
Convert to string |
to_integer |
Convert to integer |
to_boolean |
Convert to boolean |
categorical_map |
Map string categories to integers using a lookup table |
to_timestamp |
Convert to a Unix timestamp |
jsonpath |
Extract value using JSONPath |
json_parse |
Parse a JSON string into a structured value |
The categorical_map transform is particularly useful because VBL is integer-only. It maps string values (like "succeeded", "failed") to integers (like 1, 0). If a value does not match the mapping, the gather fails closed: no evidence is produced.
Webhook Ingestion
Section titled “Webhook Ingestion”Connectors can receive evidence via inbound webhooks. Each webhook connector verifies the signature of incoming payloads using the provider’s signing method:
| Method | Description |
|---|---|
STRIPE_V1 |
Stripe webhook signature verification |
GITHUB_HMAC_SHA256 |
GitHub HMAC-SHA256 webhook signatures |
VBOS_HMAC_SHA256 |
Platform HMAC for providers without a signing scheme |
Verified webhooks produce evidence with authenticated_provider_push acquisition class.
Auto-Evaluate
Section titled “Auto-Evaluate”Connectors can be configured to automatically evaluate gathered evidence against a boundary. When auto-evaluate is enabled, a successful gather immediately triggers an evaluation: no separate API call needed.
Next Steps
Section titled “Next Steps”- Authentication Patterns: connector authentication in detail
- Evidence Mapping: mapping and transform reference
- Connect Evidence Sources: step-by-step guide
