---
title: "Backfill"
method: POST
path: "/v1/embeddings/backfill"
tags: ["embeddings"]
---

# Backfill

`POST /v1/embeddings/backfill`

## Request

- Base URL: `https://api.june.januraine.ai` (the document declares no server; this is the origin it was published from)
- URL: `https://api.june.januraine.ai/v1/embeddings/backfill`
- Auth: none declared

Embed a bounded SLICE and report what remains, so the caller can loop and
show progress. Body is optional — a bodiless POST (the shape the bench client
and older callers send) still works and takes the defaults.

## Headers

- `X-Canvas` string, nullable
- `X-API-Key` string, nullable
- `Authorization` string, nullable

## Request body

- BackfillIn: One resumable slice of work. Module-level so FastAPI resolves the postponed annotation (the L2 trap: a locally-defined model under future-annotations 422s silently).
  - `max_nodes` integer
  - `batch_size` integer
  - `only_missing` boolean

## Response `200`

Successful Response

- object

## Other responses

- `422`: Validation Error

---

[API](https://skmtc.dev/januraine/apis/june-ai.md) · [All operations](https://skmtc.dev/januraine/apis/june-ai/llms.txt) · [OpenAPI document](https://skmtc.dev/januraine/apis/june-ai/revisions/097b95b163c7?raw)