Read indexed content¶
Search returns matching documents and snippets. Use the read commands when you want the indexed document text. For extracted formats such as HTML and digital PDFs, this is the canonical text stored in the index rather than raw file bytes.
List files in a collection¶
Use a prefix to narrow the listing to one part of the collection.
Read one file¶
Use a collection-relative path:
Or use a document id from search output:
Read a line range with --offset and --limit:
Read several files¶
Use multi-get when an agent or script needs several documents in one call:
Set limits when you need bounded output:
When to use each command¶
| Need | Command |
|---|---|
| See what is indexed | kbolt ls |
| Read one document | kbolt get |
| Read several documents with output limits | kbolt multi-get |
| Let an agent call read tools directly | kbolt mcp |
Next steps¶
- Use these commands from agents with Use with Claude Desktop.
- See exact arguments in Read and integration.
- Check indexed content with Health and status.