---
title: "Get Attachment Content"
method: GET
path: "/v1/templates/{template_id}/attachments/{attachment_id}/content"
tags: ["templates"]
---

# Get Attachment Content

`GET /v1/templates/{template_id}/attachments/{attachment_id}/content`

The attachment's CURRENT working text — what's actually indexed and served
to the agent — as opposed to `file_url`, which always points at the original
upload and never reflects edits (corrections, open-question answers, and the
audit loop all write to `extracted_text`, not the source file). This is the
only way for a template author to visually verify an edit landed: downloading
`file_url` shows the pristine original forever, by design (see
`attachment_processor` / `document_indexer` — the S3 object is provenance,
`extracted_text` is the live document).

## Path parameters

- `template_id` string, required
- `attachment_id` string, required

## Response `200`

Successful Response

- unknown

## Other responses

- `422` — Validation Error

---

[API](https://skmtc.dev/roxels/apis/interviewer.md) · [All operations](https://skmtc.dev/roxels/apis/interviewer/llms.txt) · [OpenAPI document](https://skmtc-service-production.skmtc.workers.dev/v1/apis/roxels/interviewer/revisions/4ef48ed226e1/schema)