FAQ
Do I need to install anything?
Section titled “Do I need to install anything?”No. Typelets runs entirely in the browser - editing, the terminal, running code, and video. Any modern desktop browser works; phones get a read-only spectator view.
Is my data saved if the container stops?
Section titled “Is my data saved if the container stops?”Your workspace files are durable - they are stored independently of any running sandbox and rebuilt into it on the next start. Anything written only inside the running sandbox at runtime (a local database file, scratch state) is not durable across a restart. Use an external store for data that must survive. See Core concepts.
Does the preview container really stay up?
Section titled “Does the preview container really stay up?”With persistent preview on, yes - it opts out of idle eviction and stays warm while the workspace exists. Without it, a sandbox is idle-evicted after about 15 minutes.
Can I run a database next to my app?
Section titled “Can I run a database next to my app?”SQLite or an in-process store works out of the box. A separate database server process can run, but its data lives in the sandbox and does not persist across a restart. For durable data, connect to an external database.
Are interviews recorded with video?
Section titled “Are interviews recorded with video?”No. Recordings capture the code timeline and Run invocations, not video. The video call is live only.
Does Typelets auto-score candidates?
Section titled “Does Typelets auto-score candidates?”Live interviews: no. The interviewer scores against the rubric. MCP can help them reason about a score; it does not write one.
Take-home assessments: yes. On submit - and when a timed assessment hits its deadline - Typelets autogrades the clone (tests when present, then Draft with AI auto-applied onto the scorecard) and emails reviewers the weighted percent, optional test counts, and notes. Reviewers can edit every chip afterward. Candidates never see the rubric or the numbers.
What can a candidate see?
Section titled “What can a candidate see?”The prompt, their own files, and public test cases. Never the rubric, hidden test expected-output, the reference solution, or scores. When a live interview ends, the candidate loses access. Take-home candidates find their work on the Candidate Dashboard in the app menu; after autograde that item is treated as scored, still without numbers. See Sharing & roles.
Where do finished take-homes go?
Section titled “Where do finished take-homes go?”In the workspace switcher, under Past assessments (muted, same pattern as Past interviews). For reviewers, the source moves there once every accepted clone is submitted and graded - submit alone is not enough. Candidates see their own clone under Past assessments as soon as they submit. See Reviewing submissions.
Is hosting available everywhere?
Section titled “Is hosting available everywhere?”Preview/hosting runs on the hosted platform. The self-hosted local Docker backend does not serve previews.
How do I automate Typelets?
Section titled “How do I automate Typelets?”Create a personal access token and use the REST API or the MCP server.
What languages can I run?
Section titled “What languages can I run?”Node, TypeScript, Python, Ruby, Java, Go, and C/C++ are installed. The Run button handles single-file entrypoints; multi-file projects run from the terminal. See Languages & runtimes.