Skip to content

he feed

Add documents to an existing knowledge abstract incrementally.

Synopsis

he feed KA_PATH INPUT [OPTIONS]

Arguments

Argument Description
KA_PATH Path to existing knowledge abstract directory
INPUT Input file, directory, or - for stdin

Supported suffixes: .txt/.md always; PDF, Word, PowerPoint, Excel, HTML, CSV, JSON, XML, EPUB and more via the optional ingest extra — pip install "hyperextract[ingest]" (see he parse for the full table). Scanned (image-only) PDFs have no text layer — run OCR first. Stdin (-) is not suffix-checked.

Options

Option Short Description
--template -t Override template (uses metadata if omitted)
--lang -l Override language (uses metadata if omitted)
--source — Source attribution (document id) — enables per-document rollback via he remove --document

Description

The feed command adds new documents to an existing knowledge abstract without losing existing data:

  1. Loads existing knowledge — Reads current knowledge abstract state
  2. Extracts from new document — Processes the new content
  3. Merges intelligently — Combines new and existing data, handling duplicates
  4. Updates metadata — Records the update timestamp

This is ideal for: - Building knowledge abstracts over time - Adding updates to existing documents - Combining information from multiple sources

Examples

Basic Usage

# Initial extraction
he parse tesla_bio.md -t general/biography_graph -o ./tesla_kb/ -l en

# Add more content
he feed ./tesla_kb/ tesla_inventions.md

Feed Multiple Documents

he feed ./ka/ doc1.md
he feed ./ka/ doc2.md
he feed ./ka/ doc3.md

Or feed a directory (same as he parse): each supported file is ingested, attributed by file stem unless --source is set. Unsupported extensions are skipped with a warning; if the directory has no readable files, the command exits non-zero.

he feed ./ka/ ./updates/

Or use a loop:

for file in updates/*.md; do
    he feed ./ka/ "$file"
done

From Stdin

cat new_content.md | he feed ./ka/ -

Merge Behavior

The merge process handles:

Scenario Behavior
Same entity Merged, descriptions combined
Same relation Updated with latest information
New entities Added to knowledge abstract
New relations Added connecting existing/new entities

Source Attribution

Pass --source with a document id to attribute the document you are feeding:

he feed ./ka/ doc1.md --source doc-1

Attribution records the document in the KA's source ledger together with a content hash of the input, which enables:

  • Change detection — feeding the same source again with unchanged content prints unchanged (content hash matches) — nothing to do and skips entirely: zero LLM calls. Use --refeed to force re-ingestion:

    he feed ./ka/ doc1.md --source doc-1 --refeed

  • Per-document rollback — everything this document contributed can later be rolled back with he remove --document doc-1

Inspect the ledger with he info --sources or ka.sources() in Python. See the Source Attribution & Provenance guide for the full picture.

Workflow Example

Building a Research Knowledge Abstract

# Day 1: Initial paper
he parse paper_v1.md -t general/concept_graph -o ./research_kb/ -l en
he show ./research_kb/

# Day 7: Updated version
he feed ./research_kb/ paper_v2.md
he show ./research_kb/

# Day 14: Related work
he feed ./research_kb/ related_work.md
he build-index ./research_kb/
he talk ./research_kb/ -q "What are the key concepts across all papers?"

Incremental Biography

# Start with early life
he parse early_life.md -t general/biography_graph -o ./bio_kb/ -l en

# Add career period
he feed ./bio_kb/ career.md

# Add later years
he feed ./bio_kb/ later_years.md

# Final visualization
he show ./bio_kb/

Verification

Check that the feed worked:

he info ./ka/

Look for: - Increased node count - Increased edge count
- Updated timestamp

Best Practices

  1. Use same template — Feeding should use compatible templates
  2. Match language — Use consistent language for best results
  3. Rebuild index after — he build-index ./ka/ for search/chat
  4. Visualize changes — he show ./ka/ to see updates

Error Handling

"Not a valid Knowledge Abstract directory"

The directory doesn't contain a valid knowledge abstract. Check:

ls ./ka/
# Should contain: data.json, metadata.json

"Template mismatch"

Feeding works best with the same template type. Override if needed:

he feed ./ka/ doc.md -t general/biography_graph

See Also

  • he parse — Create new knowledge abstract
  • he info — View knowledge abstract statistics
  • he build-index — Rebuild search index after feeding

See Also