schema-bulk

Extracts schemas from multiple files at once using a glob pattern or directory path. Either extracts distinct unique schemas (--mode distinct, default) or one schema per file (--mode perfile).

The CLI command is schema-bulk (hyphen), not schema_bulk.

--autodoc uses the same five providers as analyze: openai, openrouter, ollama, lmstudio, perplexity.

# Distinct schemas across all CSV files in a directory
undatum schema-bulk "data/*.csv" --output schemas/

# One schema per file, JSON Schema format
undatum schema-bulk data/ --mode perfile --format jsonschema --output schemas/

# With AI-powered field documentation
undatum schema-bulk "data/*.jsonl" --autodoc --output schemas/

Key options: --mode distinct|perfile, --format (yaml, json, cerberus, jsonschema, avro, parquet), --output (directory), --autodoc, --lang, --engine auto|duckdb|iterable. See schema for single-file extraction.