Skip to main content
Assets are files in your Chief project knowledge base — the same Library end users see in the app. The public API uses a three-step upload: mint an upload slot, PUT bytes to blob storage, then complete ingest on the server.

Uploaded assets land in your project's Library in Chief.

​
Lifecycle overview

​
Status values

The app shows a finer-grained version of the same progression — see Viewing asset details & file processing status if you are comparing what a user sees against what the API reports.

​
Step 1 — Create upload slot

POST /v1/assets with filename and mime_type:
Example response:

​
Step 2 — Upload bytes

PUT the file to upload_url. Apply every header in upload_headers exactly as returned. The URL expires at expires_at.

​
Step 3 — Complete ingest

POST /v1/assets/{id}/complete with an empty body. The server reads size and MD5 from blob storage and starts ingest.
Poll GET /v1/assets/{id} until status is ready or failed.

​
List assets

GET /v1/assets lists assets in the project with the same cursor pagination as chats (after_id, before_id, limit).

​
Use assets in chats

After an asset is ready, pass its id in chat scope.asset_ids so the assistant can consult that file. See Chats.

​
Labels

Attach labels with POST /v1/assets/{id}/labels. See Labels.

​
FAQ

Until the expires_at timestamp returned with the slot. If it lapses before you finish the PUT, create a new upload slot and start again.
Yes. Apply each header in upload_headers exactly as returned. Blob storage rejects the PUT if they do not match.
A file is only consultable once GET /v1/assets/{id} reports status of ready. Before that it moves through uploaded and ingesting.
POST /v1/assets/{id}/complete takes an empty body. The server reads the object’s size and MD5 from blob storage and hands the file to the ingest pipeline.
Once the asset is ready, pass its id in scope.asset_ids on a chat create or send. See Chats.
See the API reference tab under Assets, or Introduction for authentication headers.