Phantom Speculator
Automatic schema detection and inference. Analyzes payload structure to infer field types, nullability, and constraints without any configuration.
What It Does
Phantom Speculator is the zero-configuration schema inference engine. It analyzes incoming payloads and automatically infers the schema — field names, data types, nullability, cardinality, value distributions, and constraints — using statistical sampling for large payloads and producing JSON Schema-compatible output.
Capabilities
- Zero-config schema inference
- Statistical sampling for large payloads
- Nested structure detection
- Type inference (string, number, boolean, date, UUID, email)
- Nullability and cardinality analysis
- Pattern detection (email, URL, UUID, date)
- JSON Schema output
Example
const result = await mcp.call('phantom_speculator_detect', {
payload: normalizedPayload,
stream_uuid: streamUUID,
sample_size: 1000,
});
// { "type": "object", "properties": { "email": { "format": "email" } } }
console.log(result.confidence); // 0.97 Configuration
agent: phantom-speculator
version: "1.0"
inference:
sample_size: 1000
confidence_threshold: 0.95
detect_patterns: true
output:
format: json-schema
include_statistics: true Related Agents
Helix Serializer
Multi-format stream serialization. Detects and normalizes JSON, Avro, CSV, Protobuf, Parquet, and raw bytes into a canonical RedQueen envelope.
Blueprint Tracker
Schema registry and catalog. Stores, versions, and retrieves schema definitions for all data sources in the pipeline.
Solicitor
Data contract generation. Automatically generates YAML data contracts from inferred schemas, defining SLAs, quality expectations, and ownership.