Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/TrinaxCode/TrinaxAI/llms.txt

Use this file to discover all available pages before exploring further.

Collections are named RAG namespaces that let you partition your indexed content into separate, searchable buckets — for example, one collection per client, project, or topic. Every TrinaxAI instance always has at least one collection: General (id: "default"), which cannot be deleted. Collection IDs are automatically slugified via sanitize_collection_id when created.
GET /collections is open (no auth required). All write operations — POST, PATCH, and DELETE — require authorization via localhost/LAN access or an X-Admin-Token header.

​
GET /collections

List all RAG collections. This endpoint is always open and does not require authorization. The default "General" collection is always present.

​
Response

​
ok
boolean
Always true on success.
​
collections
array
Ordered list of collection objects.

​
Example

curl http://localhost:3333/collections
Response
{
  "ok": true,
  "collections": [
    {
      "id": "default",
      "name": "General",
      "created_at": 1718000000.0,
      "updated_at": 1718000000.0
    },
    {
      "id": "my-project",
      "name": "My Project",
      "created_at": 1718100000.0,
      "updated_at": 1718200000.0
    }
  ]
}

​
POST /collections

Create a new collection. The name is slugified to generate a unique id. If the slug is already taken, a numeric suffix is appended (e.g. "my-project-2").
Requires authorization.

​
Request body

​
name
string
required
Human-readable display name for the new collection. Truncated to 80 characters. The collection ID is derived by slugifying this name via sanitize_collection_id.

​
Response

​
ok
boolean
Always true on success.
​
collection
object
The newly created collection object with id, name, created_at, and updated_at.

​
Example

curl -X POST http://localhost:3333/collections \
  -H "Content-Type: application/json" \
  -H "X-Admin-Token: your-token" \
  -d '{"name": "My Project"}'
Response
{
  "ok": true,
  "collection": {
    "id": "my-project",
    "name": "My Project",
    "created_at": 1718100000.0,
    "updated_at": 1718100000.0
  }
}

​
PATCH /collections/{collection_id}

Rename an existing collection. Only the display name can be changed — the id (slug) is immutable after creation.
Requires authorization.

​
Path parameters

​
collection_id
string
required
The slug ID of the collection to rename (e.g. "my-project").

​
Request body

​
name
string
required
The new display name. Truncated to 80 characters.

​
Response

​
ok
boolean
Always true on success.
​
collection
object
The updated collection object with the new name and refreshed updated_at.

​
Example

curl -X PATCH http://localhost:3333/collections/my-project \
  -H "Content-Type: application/json" \
  -H "X-Admin-Token: your-token" \
  -d '{"name": "New Name"}'
Response
{
  "ok": true,
  "collection": {
    "id": "my-project",
    "name": "New Name",
    "created_at": 1718100000.0,
    "updated_at": 1718300000.0
  }
}

​
DELETE /collections/{collection_id}

Delete a collection and permanently remove all of its indexed chunks from the vector store. The in-memory retriever is rebuilt automatically after deletion.
This operation is irreversible. All indexed content in the collection is deleted from the LlamaIndex docstore and the on-disk manifest. You will need to re-upload and re-index to restore it. The "default" collection cannot be deleted.
Requires authorization.

​
Path parameters

​
collection_id
string
required
The slug ID of the collection to delete. Must not be "default".

​
Response

​
ok
boolean
Always true on success.
​
deleted_nodes
integer
Number of index chunks (nodes) that were removed from the vector store.

​
Example

curl -X DELETE http://localhost:3333/collections/my-project \
  -H "X-Admin-Token: your-token"
Response
{
  "ok": true,
  "deleted_nodes": 312
}

​
POST /documents/extract

Extract plain text from a PDF, DOCX, or text-based file for temporary analysis only — the content is not indexed or persisted. Useful for on-the-fly document inspection in the PWA.
Supported formats: .pdf (via pypdf, with optional OCR fallback when TRINAXAI_OCR=1), .docx (via docx2txt), .txt, .md, .mdx, .rst, .csv, .json, .xml, .yml, .yaml, .toml, .ini, .log. Maximum upload size is 15 MB (configurable via TRINAXAI_DOC_EXTRACT_MAX_BYTES). Text is truncated at 120,000 characters (configurable via TRINAXAI_DOC_EXTRACT_MAX_CHARS).

​
Request

multipart/form-data with a single file field.
​
file
file
required
The document to extract text from. Must be one of the supported formats listed above.

​
Response

​
ok
boolean
Always true on success.
​
name
string
The original filename of the uploaded document.
​
text
string
The extracted plain text. Truncated to DOC_EXTRACT_MAX_CHARS (default 120,000) if the document is very long.
​
chars
integer
Character count of the (possibly truncated) returned text.
​
truncated
boolean
true if the extracted text exceeded the character limit and was cut off.

​
Example

curl -X POST http://localhost:3333/documents/extract \
  -F "file=@/path/to/report.pdf"
Response
{
  "ok": true,
  "name": "report.pdf",
  "text": "[Page 1]\nQuarterly Report Q2 2024\n\nThis report covers...",
  "chars": 45200,
  "truncated": false
}

Build docs developers (and LLMs) love

Get started for freeTalk to us