
Batch-get collection document metadata
Batch-get collection document metadata
Fetch metadata for many known file IDs in one Management API call with GET https://management-api.x.ai/v1/collections/{collection_id}/documents:batchGet. Use this when you already have IDs (from an upload receipt or your own store) and want status without paging the full list. Listing and single-doc poll: List and poll documents in a Grok collection.
Batch get
curl "https://management-api.x.ai/v1/collections/$COLLECTION_ID/documents:batchGet?file_ids=$FILE_ID_1&file_ids=$FILE_ID_2" \
-H "Authorization: Bearer $XAI_MANAGEMENT_API_KEY"
file_ids is required (repeat the query param once per id). Optional team_id when the key can see more than one team; otherwise the team comes from the credentials.
import os
import requests
headers = {"Authorization": f"Bearer {os.environ['XAI_MANAGEMENT_API_KEY']}"}
collection_id = "collection_dbc087b1-6c99-493d-86c6-b401fee34a9d"
file_ids = [
"file_55a709d4-8edc-4f83-84d9-9f04fe49f832",
"file_94847856-a56f-4b1e-82dd-7fe0b3af43d9",
]
params = [("file_ids", fid) for fid in file_ids]
resp = requests.get(
f"https://management-api.x.ai/v1/collections/{collection_id}/documents:batchGet",
headers=headers,
params=params,
)
resp.raise_for_status()
for doc in resp.json().get("documents", []):
meta = doc.get("file_metadata") or {}
print(meta.get("file_id"), doc.get("status"), doc.get("error_message"))
Each document includes file_metadata (file_id, name, size_bytes, content_type, timestamps), optional fields, status, and error_message. Statuses match the single-get path: DOCUMENT_STATUS_PROCESSING, DOCUMENT_STATUS_PROCESSED, DOCUMENT_STATUS_FAILED (plus unknown).
Pitfalls
- Batch get does not invent IDs. Pass the
file_idvalues you already stored. - Searching while status is still processing returns empty matches — poll until
DOCUMENT_STATUS_PROCESSED. - Management key for list/get/batchGet/delete.
/v1/documents/searchusesXAI_API_KEY. - Console API credits are a separate bill from SuperGrok's weekly pool on grok.com.