Admin Tool

The Traverse Admin tool (traverse-admin) provides offline bulk import from CSV and Cypher files, plus update management.

CSV Import

Import CSV files into a new .tvdb database. Specify node files with --nodes LABEL=file.csv and edge files with --edges TYPE=file.csv:

traverse-admin import \
  --nodes Person=people.csv \
  --nodes Company=companies.csv \
  --edges WORKS_AT=employment.csv \
  -o ./data/company.tvdb

Options

Flag	Short	Description	Default
`--nodes LABEL=file.csv`	—	CSV file for nodes with the given label (repeatable)	—
`--edges TYPE=file.csv`	—	CSV file for edges with the given type (repeatable)	—
`--output <PATH>`	`-o`	Output `.tvdb` file path	—
`--id-type <string\|int>`	—	ID column type for cross-referencing	`string`
`--delimiter <CHAR>`	—	CSV field delimiter (single character)	`,`
`--skip-bad-rows`	—	Skip rows that fail to parse instead of aborting	—

CSV Format

Node CSV files must include a header row. The :ID column is used as the node identifier for edge resolution. Columns starting with : (Neo4j meta columns) are skipped:

:ID,name,age
1,Alice,30
2,Bob,25
3,Carol,28

Edge CSV files must include :START_ID and :END_ID columns that reference node IDs:

:START_ID,:END_ID,since
1,2,2020
2,3,2021

Type Inference

Property values are auto-inferred from CSV data:

true / false → boolean
Integer values → 64-bit integer
Decimal values → 64-bit float
Empty cells → null (skipped)
Everything else → string

Cypher Import

Import a file of Cypher statements (one per line) into a new .tvdb database:

traverse-admin cypher-import \
  -i data.cypher \
  -o ./data/mydb.tvdb

Options

Flag	Short	Description	Default
`--input <FILE>`	`-i`	Cypher file (one statement per line)	—
`--output <PATH>`	`-o`	Output `.tvdb` file path	—
`--skip-errors`	—	Continue on statement errors	—
`--index LABEL:PROP`	—	Pre-create an index (repeatable)	—

File Format

One statement per line (trailing semicolons are stripped)
Empty lines and // comment lines are skipped
Multi-line statements continue until ; or end-of-file

Auto-Indexing

When the first MATCH statement is encountered (after a series of CREATE statements), Traverse auto-creates indexes on all :Label {prop} patterns seen in preceding CREATEs. This avoids O(N) scans for edge creation.

You can override this with explicit --index flags, which are created before any Cypher statements execute:

traverse-admin cypher-import \
  -i data.cypher \
  -o output.tvdb \
  --index Person:id \
  --index Company:name

Tip: Pre-creating indexes with --index significantly speeds up imports that rely on MERGE or property lookups.

Update Checking

Check if a newer version of Traverse is available:

traverse-admin update

Download the latest version to the current directory:

traverse-admin update --download