Skip to main content
POST
/
v1
/
ingest
Ingest
curl --request POST \
  --url https://api.userepo.com/v1/ingest \
  --header 'Content-Type: application/json' \
  --data '
{
  "documents": [
    {}
  ],
  "connectorId": "<string>",
  "provider": "<string>",
  "sourceKind": "<string>",
  "externalId": "<string>",
  "title": "<string>",
  "body": "<string>",
  "url": "<string>",
  "authorName": "<string>",
  "sourceCreatedAt": {},
  "sourceUpdatedAt": {},
  "metadata": {}
}
'

Documentation Index

Fetch the complete documentation index at: https://docs.userepo.com/llms.txt

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

Use this when you have a custom source not covered by Repo’s built-in connectors (e.g. an internal wiki, a proprietary CRM, a folder of PDFs in S3) and you want to feed its documents into Repo’s memory.

​
Required action

ingest

​
Body

​
documents
array
required
Array of source documents to ingest. Min 1 entry.

​
Document fields

​
connectorId
uuid
required
The connector this document belongs to. Create a connector first via the console, or contact support for a “virtual” connector for custom sources.
​
provider
string
required
One of slack, google_drive, notion, gmail. For custom sources, pick the one whose access model matches yours.
​
sourceKind
string
required
One of message, document, file.
​
externalId
string
required
Stable ID from your source system. Repo dedupes on (organizationId, provider, externalId) — re-posting with the same externalId is treated as an update.
​
title
string
required
Human-facing title. Used in citations.
​
body
string
required
The full text content to embed.
​
url
string
Deep-link back to the source (will be returned in retrieval hits).
​
authorName
string
Original author’s name.
​
sourceCreatedAt
ISO 8601
When the source was created at the provider.
​
sourceUpdatedAt
ISO 8601
When the source was last updated at the provider.
​
metadata
object
default:"{}"
Arbitrary provider-specific metadata. If you want to preserve ACLs or other access policy data, include an accessPolicy key — Repo’s retrieval will surface it back to the caller.

​
Response

{
  "ingested": 12,
  "skipped": 0
}
skipped counts documents that matched an existing externalId with identical content (no-op).

​
Example

curl -X POST https://api.userepo.com/v1/ingest \
  -H "Authorization: Bearer repo_your_key" \
  -H "Content-Type: application/json" \
  -d '{
    "documents": [
      {
        "connectorId": "550e8400-e29b-41d4-a716-446655440000",
        "provider": "notion",
        "sourceKind": "document",
        "externalId": "internal-wiki-page-42",
        "title": "Refund policy",
        "body": "Customers can request a refund within 30 days...",
        "url": "https://wiki.acme.internal/refunds",
        "metadata": {
          "accessPolicy": {
            "provider": "notion",
            "mode": "provider_scope",
            "description": "Wiki page accessible to all employees"
          }
        }
      }
    ]
  }'

​
When to use ingest vs. a connector

Use /v1/ingest whenUse a built-in connector when
You have a custom source not covered by connectorsThe source is Slack, Drive, Notion, or Gmail
You want to control sync cadence yourselfYou want hands-off, every-15-minute polling
You’re piping in data from an ETL pipeline you already runYou don’t want to maintain sync logic