For the complete documentation index, see llms.txt. This page is also available as Markdown.

Files & configurations

Every project is also a file tree — stored content layered over the Playwright template — and you can edit the safe parts of it.

Each project has a file tree that the sandbox materializes for every generation and run. It has two layers:

Source
What it is
Writable via API?

project

Files stored in TestVibe: your .feature sources, the AI-generated specs, your custom configs.

Partly — see the allowlist below.

template

The Playwright scaffolding every project starts from (playwright.config.js, helpers, Configurations/Default.config.js).

No — read-only fallback.

Reads check stored content first and fall back to the template, so read_file always shows what the next run will actually use.

File kinds

Files are classified by path: feature (Gherkin under Features/), spec (generated tests), config (under Configurations/), and supporting files.

The write allowlist

write_file is deliberately narrow. A write is accepted only if:

  • the path is under Features/ or Configurations/,

  • it contains no .. segments,

  • the content is ≤ 256 KB.

Everything else — generated specs, the template scaffolding — is owned by the generation pipeline and rejected with 400 invalid_request.

Writing a new configuration? Read Configurations/Default.config.js first and use it as the reference shape; the guide walks through a Mobile config end to end.

Written files take effect on the next run or generation — there's no restart or publish step.

Last updated

Was this helpful?