R Rubric

FAQ

Quick answers to questions that come up often.

Where do I get an API key?

  • Anthropic: sign in at the Anthropic console, go to API keys, create one. Format: sk-ant-....
  • OpenAI: sign in at the OpenAI dashboard, create a new key. Format: sk-....

You only need a key from the provider you intend to use. Switching providers later just means pasting a different key in Settings.

Is my interview data sent anywhere?

The minimum-trust answer: only when you trigger an AI feature, and only to the configured provider.

  • Sessions, scores, and notes live on your machine.
  • Generated interviews go to your configured storage backend (local disk, your git repo, or your S3 bucket — all under your control).
  • AI calls (interview generation, report drafting) send the relevant payload directly to Anthropic or OpenAI. Rubric runs no servers and proxies no requests.
  • If you leave the API key blank, no network calls are made for interview content.

Why are rubric categories weighted?

Because not every dimension matters equally for every role. A senior architect interview probably weights “Trade-off Reasoning” heavier than “Communication”; a tech-lead interview probably reverses that. Weights let one rubric template be honest about what’s being measured for this role, in this round.

If you don’t want weights, set every category to weight 1 — the math collapses to a plain average.

Can I bring my own interviews?

Yes — use the Generate screen. It produces a complete interview package (candidate board, interviewer guide, weighted rubric) from a one-line role prompt, plus optional cloud and difficulty hints. The generated interview is added to your library right away, and the source files are saved to your configured storage backend.

If you have direct filesystem access to your storage backend (the local-filesystem option), the generated interviewer-guide.md and interview.yaml files are plain text and can be hand-edited between rounds. The bundled starter catalog is read-only — you add to it via Generate, you don’t replace it.

Can I export a single report?

Yes — see Interview Reports. A finished report exports as Markdown (paste-ready) or PDF (styled with the Rubric brand).

If you want to export a session’s raw data (scores, notes, timestamps) for analysis elsewhere, that’s stored as JSON in your configured storage backend’s sessions/ directory.

What platforms is the app supported on?

Rubric ships signed installers for:

  • macOS — primary target, most polished. Apple Silicon and Intel.
  • Windows — 64-bit signed installer.
  • Linux — portable AppImage.

The four-panel live-session layout is designed for a desktop-class window of at least 1280 × 800.

Where is configuration stored?

In a Hive database in the platform’s standard application-support directory:

  • macOS: ~/Library/Application Support/rubric/
  • Windows: %APPDATA%\rubric\
  • Linux: ~/.local/share/rubric/

API keys, storage backend config, and provider/model selection live there. To wipe configuration, delete that directory and restart the app.

Why isn’t there a dark mode?

The cream / ink letterpress design is a deliberate aesthetic choice, not a default to be flipped. If a dark mode would meaningfully change how you use Rubric, send the request through the support channel.

Can I run interviews offline?

Yes. The Live Session, scoring, notes, and the bundled interview library are all fully offline. The two operations that require network are interview generation and report drafting, both of which call your configured AI provider.

If you finish a session offline, just save it. Open it later when you’re online and click Regenerate to draft the report.

How do I report a bug or send feedback?

Use the support channel. Useful things to include in a bug report:

  • What you were doing when it happened (which screen, which interview).
  • What you expected vs. what actually happened.
  • Your platform (macOS / Windows / Linux) and Rubric version (shown on the Help screen).