Evidence intelligence

ContractForge evidence is a core model. Adapters decide where it is stored:

Platform	Typical evidence store
Databricks	Delta control tables in a catalog/schema.
AWS	Iceberg tables in Glue Catalog queried through Athena or Glue Spark.
Future adapters	Platform-native audit/evidence tables.

ContractForge AI analyzes the normalized evidence model, not one platform's runtime.

Supported evidence sections

The analyzer accepts:

• runs;
• errors;
• quality;
• quarantine;
• schema changes;
• lineage;
• streams;
• annotations;
• access;
• operations;
• cost;
• state;
• locks;
• collection errors.

Databricks table-name aliases such as ctrl_ingestion_runs are accepted. AWS aliases such as aws_runs, aws_quality, aws_state and aws_cost are also accepted.

Databricks input

{
  "scope": {
    "platform": "databricks",
    "catalog": "workspace",
    "ctrl_schema": "contractforge_ops",
    "target_table": "workspace.bronze.b_products"
  },
  "ctrl_ingestion_runs": [],
  "ctrl_ingestion_quality": [],
  "ctrl_ingestion_state": []
}

AWS input

{
  "scope": {
    "platform": "aws",
    "evidence_store": "iceberg",
    "database": "contractforge_ops",
    "target_table": "glue_catalog.analytics.b_products"
  },
  "aws_runs": [],
  "aws_quality": [],
  "aws_state": []
}

Command

contractforge-ai analyze-control-tables \
  --input evidence.json \
  --format html

The deterministic analyzer reports failure clusters, quality and quarantine volume, schema drift, stream inconsistencies, governance application failures, stale state, missing cost evidence and partial evidence coverage. Provider enrichment may explain remediation, but it cannot change deterministic risk or status.