🌐 Veritas Engine API (DSE)

Last Updated: March 23, 2026 12:00 UTC

LIVELLO 2 service for signal quality validation, regime detection, and confidence aggregation.

📍 Location

Service: services/api_veritas_engine/
Pure core: vitruvyan_core/core/governance/veritas_engine/
Module doc: Veritas Engine

🚪 Base URL / Port

Config default: SERVICE_PORT=8013 (services/api_veritas_engine/config.py)
Docker external port: 9013 (see infrastructure/docker/docker-compose.yml)

🧭 Endpoints (implemented)

Defined in services/api_veritas_engine/api/routes.py:

Method	Path	Description
`GET`	`/health`	liveness + bus connectivity status
`POST`	`/validate`	run a full signal quality validation
`GET`	`/validate/{run_id}`	retrieve a past run (501 — not yet implemented)

🧾 Request/Response models

See services/api_veritas_engine/models/schemas.py.

`POST /validate`

Request (ValidationRunRequest):

{
  "entity_ids": ["AAPL", "MSFT", "GOOGL"],
  "features": {
    "AAPL": {
      "completeness": 0.95,
      "z_consistency": 0.88,
      "freshness": 0.90,
      "sample_adequacy": 0.75,
      "outlier_probability": 0.05,
      "value": 185.5,
      "staleness_days": 0
    }
  },
  "domain": "finance",
  "diagnostics_level": "standard"
}

Response (ValidationRunResponse):

{
  "run_id": "ve-20260323-a1b2c3d4",
  "profile": "default",
  "entity_count": 3,
  "overall_confidence": 0.872,
  "regime": {
    "classification": "normal",
    "score": 0.21,
    "components": {}
  },
  "qualities": [
    {
      "entity_id": "AAPL",
      "quality_score": 0.91,
      "flags": [],
      "components": {}
    }
  ],
  "produced_at": "2026-03-23T12:00:00Z",
  "diagnostics": {}
}

`GET /health`

Response (HealthCheckResponse):

{
  "status": "ok",
  "service": "api_veritas_engine",
  "bus_connected": true
}

⚙️ Env vars (service-level)

Loaded in services/api_veritas_engine/config.py:

Variable	Default	Description
`SERVICE_PORT`	`8013`	HTTP listen port
`LOG_LEVEL`	`INFO`	logging verbosity
`QUALITY_PROFILE`	`default`	threshold preset (`default`/`strict`/`relaxed`)
`NEURAL_ENGINE_URL`	`http://neural_engine:8003`	upstream NE URL (optional)
`REDIS_URL`	`redis://redis:6379`	StreamBus connection
`POSTGRES_HOST`	`postgres`	audit log DB host
`POSTGRES_DB`	`vitruvyan`	audit log DB name
`POSTGRES_USER`	`vitruvyan`	audit log DB user
`POSTGRES_PASSWORD`	(secret)	audit log DB password

🔄 Redis Streams Listener

services/api_veritas_engine/streams_listener.py runs as a sidecar container.

Channel: veritas.validation.requested
Consumer group: veritas_engine
Trigger: deserializes payload → calls VeritasBusAdapter.run_validation()
ACK: acknowledges event after successful processing

🗄️ Audit Log (optional)

services/api_veritas_engine/adapters/persistence.py writes to:

TABLE veritas_audit_log (
  run_id        TEXT PRIMARY KEY,
  profile       TEXT,
  entity_count  INTEGER,
  overall_confidence FLOAT,
  regime        TEXT,
  produced_at   TIMESTAMPTZ,
  diagnostics   JSONB
)

Insert is ON CONFLICT DO NOTHING — idempotent. Postgres wiring is optional: if PostgresAgent is unavailable the audit log silently skips.

🐳 Docker

# infrastructure/docker/docker-compose.yml (excerpt)
veritas_engine:
  build: services/api_veritas_engine
  ports:
    - "9013:8013"
  environment:
    - QUALITY_PROFILE=default
    - REDIS_URL=redis://redis:6379

# Build + deploy
docker compose build veritas_engine
docker compose up -d veritas_engine

🔗 Related

Neural Engine API Horizon Engine API