Skip to main content

Request Editor

The Request Editor is VAAST's manual testing surface for AI endpoints. You build one request at a time, send it, inspect the response, and iterate. Everything you send is scoped to the current workspace and inherits the workspace's authentication configuration.

Use the Request Editor when you want to:

  • Reproduce a finding from an automated scan by hand.
  • Craft a one-off prompt-injection probe against a target.
  • Explore a new endpoint before adding it to a scan.
  • Verify that your workspace auth config actually reaches the target.

Building a request​

The editor is organised into four panels:

  1. Method and URL — pick an HTTP method (GET, POST, PUT, PATCH, DELETE) and set the URL. The workspace's target URL is used as the default.
  2. Headers — add, edit, or remove headers. The workspace auth header is added automatically and marked read-only; override it for a single request from the overflow menu.
  3. Body — the body panel supports raw text, JSON (with syntax-highlighted formatting), and form-encoded payloads. Switching the type preserves the content where it makes sense.
  4. Response — after you hit Send, the response panel shows the status code, latency, response headers, and response body. Bodies are pretty-printed for JSON and collapsible for large payloads.

Request history​

Every request you send from the editor lands in the History drawer for the current workspace. History entries include the full request/response pair and a timestamp. You can:

  • Re-open a past request to continue iterating from it.
  • Pin a request so it stays at the top of the list.
  • Clear history for the workspace. Cleared history cannot be recovered.

History is stored locally, alongside your scan history, inside the workspace. Nothing is synced off-device.

Relationship to scans​

The Request Editor is a standalone tool, not a subset of the scanner. Payloads fired from a scan do not appear in the Request Editor history, and requests you hand-craft in the editor do not feed back into the scan engine. If you find a new attack by hand and want to turn it into a repeatable check, add it to a payload library YAML file.