Reading content

Two command groups let you read beyond a search hit: mnm chunks walks the chunk graph by ID, and mnm documents gives you a document overview and a windowed view of its full chunk list.

mnm chunks — walk the chunk graph

Every search result carries a chunk_id. Use these subcommands to read around it.

mnm chunks show <chunk-id>

Fetch and render one chunk with its document and source context.

mnm chunks show 019682f0-1234-7abc-8def-0123456789ab
mnm chunks show 019682f0-1234-7abc-8def-0123456789ab --json

mnm chunks next <chunk-id>

Fetch the next N chunks after the anchor in the same document.

mnm chunks next <chunk-id>              # default: 5 chunks
mnm chunks next <chunk-id> --count 10  # up to 100 server-side
mnm chunks next <chunk-id> --full      # full content (default is a 240-char preview)

Flag	Default	Notes
--count <n>	5	Number of chunks to fetch (clamped to 1..=100 server-side).
--full	off	Show full chunk content instead of a 240-character preview.

mnm chunks prev <chunk-id>

Same flags as next, but fetches chunks before the anchor.

mnm chunks prev <chunk-id> --count 3

mnm chunks neighbors <chunk-id>

Fetch prev + anchor + next in one call. Equivalent to running prev, show, and next separately but with a single round-trip.

mnm chunks neighbors <chunk-id>
mnm chunks neighbors <chunk-id> --count 5 --full

mnm documents — read documents

Every search result also carries a document_id. Use these subcommands to understand the document's shape and read it in windows.

mnm documents show <doc-id>

Show the document overview: metadata, total chunk count, and the ordered chunk skeleton (index, heading path, and preview for each chunk). This is the right starting point when you want to understand a document's structure before reading it in depth.

mnm documents show <doc-id>
mnm documents show <doc-id> --json

mnm documents chunks <doc-id>

Render a windowed slice of the document's chunks, for reading long documents section by section.

mnm documents chunks <doc-id>                      # default: first 20 chunks
mnm documents chunks <doc-id> --from 20 --limit 20 # second window

Flag	Default	Notes
--from <n>	0	Starting chunk index offset (0-based).
--limit <n>	20	Maximum chunks to return in this window.

Typical workflow

A common pattern is to search for a topic, identify a relevant hit, then read the surrounding context:

# 1. Search — grab a chunk_id and document_id from the results
mnm search "Compact ledger Map ADT" --limit 3 --json \
  | jq '{chunk: .results[0].chunk_id, doc: .results[0].document_id}'

# 2. Follow the chunk's neighbours to read around the hit
mnm chunks next <chunk-id> --count 10

# 3. Size the document to understand how much there is
mnm documents show <doc-id>

# 4. Read window by window
mnm documents chunks <doc-id> --from 0 --limit 20
mnm documents chunks <doc-id> --from 20 --limit 20

All of these commands support --json for scripting.