Vadalog API

The Vadalog API exposes JarvisPy routes for analyzing programs, managing bind annotations, and evaluating logic against the Vadalog engine.

Python SDK: There are no dedicated Vadalog helpers in prometheux_chain. Call these endpoints with your HTTP client (for example httpx or requests) against your JarvisPy base URL, with a valid JWT.

Analyze Program

Inspect a Vadalog program without executing it. Returns predicate structure derived from the parser.

HTTP: POST /api/v1/vadalog/analyze

Body (JSON)

Field	Type	Required	Description
`program`	string	Yes	Vadalog program source
`conceptType`	string	No	When `"sql"`, `program` is treated as SQL (use with `conceptName`)
`conceptName`	string	No	Concept / output predicate name when using SQL mode

Response (data)

Field	Description
`headPredicates`	Head predicates in the program
`inputPredicates`	Input predicates

curl -X POST "https://api.prometheux.ai/jarvispy/my-org/my-user/api/v1/vadalog/analyze" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN" \
  -d '{
    "program": "@input(\"products\").\n@output(\"expensive_products\").\nexpensive_products(Product, Price) <- products(Product, Price), Price > 100."
  }'

Example shape of data:

{
  "headPredicates": ["expensive_products"],
  "inputPredicates": ["products"]
}

Parse Binds

Parse @bind / @qbind annotations from a program and separate rule code from datasource bindings.

HTTP: POST /api/v1/vadalog/parse-binds

Body (JSON)

Field	Type	Required	Description
`program`	string	Yes	Full Vadalog source including annotations
`outputPredicate`	string	No	Extra predicate name to treat as output when classifying binds

Response (data)

Field	Description
`code`	Program text without annotation lines
`binds`	`{ "input": [...], "output": [...] }` — each entry includes `predicate`, `annotation`, and parsed fields when available

curl -X POST "https://api.prometheux.ai/jarvispy/my-org/my-user/api/v1/vadalog/parse-binds" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN" \
  -d '{
    "program": "@bind(\"data\",\"csv useHeaders=true\",\"/tmp\",\"test.csv\").\nresult(X) <- data(X).\n@output(\"result\").",
    "outputPredicate": "result"
  }'

Build Bind

Rewrite an existing bind annotation for a new predicate name. When isOutput is true, the output target (e.g. CSV filename or query name) is updated for output usage.

HTTP: POST /api/v1/vadalog/build-bind

Body (JSON)

Field	Type	Required	Description
`bindAnnotation`	string	Yes	Full `@bind(...)` or `@qbind(...)` annotation
`predicateName`	string	Yes	New predicate name
`isOutput`	boolean	No	Default `false`. If `true`, adjust output table/query naming

Response (data)

Field	Description
`annotation`	Updated annotation string

curl -X POST "https://api.prometheux.ai/jarvispy/my-org/my-user/api/v1/vadalog/build-bind" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN" \
  -d '{
    "bindAnnotation": "@bind(\"old_pred\",\"csv useHeaders=true\",\"/tmp\",\"data.csv\").",
    "predicateName": "new_pred",
    "isOutput": false
  }'

Evaluate Program

Run a Vadalog program with optional parameters and compute hints. JarvisPy parses the program and delegates execution to the engine (local or configured compute).

Python SDK
REST API

Not available in SDK — use REST API directly.

There is no evaluate_vadalog (or similar) in the Python SDK. Use POST /api/v1/vadalog/evaluate with requests, httpx, or another HTTP client.

HTTP Request

POST /api/v1/vadalog/evaluate

Body (JSON)

Field	Type	Required	Description
`program`	string	Yes	Vadalog program source
`params`	object	No	Parameter map passed to evaluation (default `{}`)
`compute`	object	No	Optional compute overrides (resolved with the authenticated user's compute configuration)

cURL Example

curl -X POST "https://api.prometheux.ai/jarvispy/my-org/my-user/api/v1/vadalog/evaluate" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN" \
  -d '{
    "program": "@input(\"products\").\n@output(\"expensive_products\").\nexpensive_products(Product, Price) <- products(Product, Price), Price > ?threshold.",
    "params": {
      "threshold": 100
    }
  }'

Response

Successful responses use the standard JarvisPy envelope: status, message, and data. The data field contains the evaluation payload returned by the Vadalog client (structure depends on the program and engine).

{
  "data": {},
  "message": "Vadalog program evaluated successfully",
  "status": "success"
}

Evaluate with parameters (engine proxy)

Proxy to the Vadalog engine's evaluateWithParams endpoint. Uses multipart form data (application/x-www-form-urlencoded / multipart/form-data), not JSON.

HTTP: POST /api/v1/vadalog/evaluateWithParams

Input	Description
Form: `program`	Required. Vadalog program text
Query: `measureElapsedTime`	Optional. Default `false`

JarvisPy forwards program and measureElapsedTime to the engine and supplies empty default params and databricksConfigs in the proxied request.

The response body is whatever the engine returns (not necessarily the same envelope as POST /evaluate).

curl -X POST "https://api.prometheux.ai/jarvispy/my-org/my-user/api/v1/vadalog/evaluateWithParams?measureElapsedTime=false" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN" \
  --form-string 'program=@input("x").@output("y").y(1).'

Stop Evaluation

Signals the engine to stop the current evaluation and breaks any in-flight wait on the JarvisPy side.

HTTP: GET /api/v1/vadalog/stop

curl -X GET "https://api.prometheux.ai/jarvispy/my-org/my-user/api/v1/vadalog/stop" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN"

Get Engine Status

Lightweight JarvisPy health-style check for the evaluation service (not per-execution polling).

HTTP: GET /api/v1/vadalog/status

Response (data)

Field	Description
`status`	Service readiness indicator (e.g. `"ready"`)
`timestamp`	ISO timestamp

curl -X GET "https://api.prometheux.ai/jarvispy/my-org/my-user/api/v1/vadalog/status" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN"

Example:

{
  "data": {
    "status": "ready",
    "timestamp": "2026-03-30T12:00:00.000000"
  },
  "message": "Evaluation service status retrieved successfully",
  "status": "success"
}

Complete Workflow Example

REST-only flow: analyze structure, optionally parse binds, then evaluate.

# 1. Analyze structure
curl -X POST "https://api.prometheux.ai/jarvispy/my-org/my-user/api/v1/vadalog/analyze" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN" \
  -d '{"program": "@input(\"products\").\n@output(\"expensive_products\").\nexpensive_products(Product, Price) <- products(Product, Price), Price > ?threshold."}'

# 2. Parse binds (skip if the program has no @bind / @qbind)
curl -X POST "https://api.prometheux.ai/jarvispy/my-org/my-user/api/v1/vadalog/parse-binds" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN" \
  -d '{"program": "@input(\"products\").\n@output(\"expensive_products\").\nexpensive_products(Product, Price) <- products(Product, Price), Price > ?threshold.", "outputPredicate": "expensive_products"}'

# 3. Evaluate with parameters
curl -X POST "https://api.prometheux.ai/jarvispy/my-org/my-user/api/v1/vadalog/evaluate" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_JWT_TOKEN" \
  -d '{"program": "@input(\"products\").\n@output(\"expensive_products\").\nexpensive_products(Product, Price) <- products(Product, Price), Price > ?threshold.", "params": {"threshold": 100}}'

Replace YOUR_JWT_TOKEN and the org/user path with your values.

Analyze Program​

Parse Binds​

Build Bind​

Evaluate Program​

HTTP Request​

Body (JSON)​

cURL Example​

Response​

Evaluate with parameters (engine proxy)​

Stop Evaluation​

Get Engine Status​

Complete Workflow Example​

Analyze Program

Parse Binds

Build Bind

Evaluate Program

HTTP Request

Body (JSON)

cURL Example

Response

Evaluate with parameters (engine proxy)

Stop Evaluation

Get Engine Status

Complete Workflow Example