Documentation Index
Fetch the complete documentation index at: https://docs.knowledgestack.ai/llms.txt
Use this file to discover all available pages before exploring further.
Overview
Knowledge Stack organizes all content — folders, documents, versions, sections, and chunks — in a unified tree hierarchy. The content traversal API lets you navigate this tree with a single, flexible endpoint that works at any level of the hierarchy.
Instead of using separate endpoints for listing documents in a folder, versions of a document, or sections in a version, you use a single traversal function that starts from any point in the tree and returns its children.
How It Works
Every item in Knowledge Stack has a path part — a node in the content tree. The tree structure looks like this:
Folder
-> Document
-> Document Version
-> Section (optional)
-> Chunk
-> Thread
-> Thread Message
Common Use Cases
| What You Want | Starting Point | Depth | Type Filter |
|---|
| List documents in a folder | Folder | 1 | DOCUMENT |
| List documents and subfolders | Folder | 1 | DOCUMENT, FOLDER |
| Find all documents recursively | Folder | unlimited | DOCUMENT |
| List versions of a document | Document | 1 | DOCUMENT_VERSION |
| List sections in a version | Document version | 1 | SECTION |
| Find all chunks under a folder | Folder | unlimited | CHUNK |
| List thread messages | Thread | 1 | THREAD_MESSAGE |
| List all direct children | Any node | 1 | (none) |
Parameters
Depth
Controls how deep to traverse:
| Value | Behavior |
|---|
1 | Direct children only (most common) |
-1 | Traverse all the way to leaf nodes (capped at 30 levels) |
N | Traverse up to N levels deep (capped at 30) |
Type Filter
Filters the returned results by content type. Intermediate nodes are traversed but excluded from results unless their type matches the filter. For example, when searching for chunks under a folder, the system traverses through documents, versions, and sections but only returns chunks.
Ordering
Results can be ordered in four ways:
| Order | Description |
|---|
NAME | Alphabetical by name (default) |
LOGICAL | User-defined ordering (drag-and-drop position) |
UPDATED_AT | Most recently updated first |
CREATED_AT | Oldest first |
Include Parent
When enabled, the starting node is included in the results alongside its children. This is useful for:
- Building breadcrumbs (showing the current folder name alongside its children)
- Serializing a subtree as a self-contained structure
- Loading parent metadata in a single request
Traverse Until Types
Stops traversal at specific content types while still including them in results. For example, you can traverse folders but stop at documents — the documents are returned, but their versions, sections, and chunks are not traversed.
For large result sets, use pagination with offset and limit parameters. Without pagination, results are capped at 250 items. The response includes a total count so you can determine if more results are available.
Limits
| Limit | Value |
|---|
| Maximum traversal depth | 30 levels |
| Maximum unpaginated results | 250 items |
Edge Cases
| Scenario | Behavior |
|---|
| Starting node does not exist | Returns a 404 error |
| Starting node has no children | Returns an empty list (not an error) |
| Results exceed 250 items | Returns the first 250 with the total count. Use pagination to retrieve more. |
