Sealed — skills run sealed. The skill executes server-side; its source is never distributed.
Builds a non-leading customer-interview script from your product hypothesis: past-behavior questions that surface truth, not compliments.
$0.14 per call · runs on sonnet-tier compute
Listing updated 2026-06-12
Safety: Containment verified 2026-06-12. Safety screened under policy 2026-06-13.13 on 2026-08-24. Sealed review approved.
| Field | Description |
|---|---|
hypothesis | What you believe (who has the problem, what they'd pay for) |
audience | Who you're interviewing |
curl -X POST https://sealed.run/run/user-interview-script-builder \
-H "Authorization: Bearer sealed_sk_..." \
-H "Content-Type: application/json" \
-d '{"inputs":{"hypothesis":"...","audience":"..."}}'
// Sealed MCP server tools: list_skills, run_skill. Tool call:
run_skill({"skillId":"user-interview-script-builder","inputs":{"hypothesis":"...","audience":"..."}})
sealed run user-interview-script-builder --input hypothesis="..." --input audience="..."
To get a key and connect your agent, see Add Sealed to your agent. How the run stays sealed: server-side skill execution.
Browse all skills on the marketplace · Sell a skill like this — the buyer never gets your file
Machine-readable catalog: /skills.json