# Sandbox Gate

> Check an API platform's test mode so integrators validate every outcome before they touch real money, messages or
> data. List the published error codes and every documented outcome with the test value that triggers it; the browser
> measures failure-path coverage against the 100% gate, holds every magic value to the real error catalog, catches one
> value meaning two things, verifies the Luhn digit of every test card number, checks event and time coverage, keys,
> isolation, reset, quotas, card-testing controls and the path to live, and ranks the isolation and simulation choices
> from a short interview. Metered runs write the test-mode design, and the cutover to live. Built on the SkillSafe skill
> @samber/api-test-mode-design, from samber/developer-platform-skills (MIT), by Samuel Berthe.

https://sandbox-gate.skillsafe.ai/

## What is free
- Failure paths: every documented outcome triggerable by a magic value or a simulation endpoint (gate: 100%), failure outcomes as well as success, every code in the published catalog (never a sandbox-only error shape), one meaning per value or a label per operation, a valid Luhn check digit on every test card number.
- Events and time: every async event fireable on demand (gate: 100%), CLI events signed or documented, every time-dependent flow testable by a test clock or auto-aging fixtures rather than waiting, synthetic seed data, reset endpoints and periodic cleanup.
- Isolation and keys: a stateful test mode rather than a mock, the isolation level stated in the docs, a mirrored end-user surface, configuration from production's infrastructure-as-code, a type prefix plus _test_ or _live_, a named wrong-environment error, secret hygiene for test keys, production limits or an aggregate cap for sandbox quotas, card-testing controls on live checkout.
- The path to live: an explicit environment-not-approved error, the four-step cutover checklist with a live canary, the five pitfalls published.
- The rankings: isolation (soft toggle, separate copy, integrator convention) and simulation (magic values, simulation endpoints, fixture builder).
- A Luhn tool, the test-value catalog as Markdown, and exports: the checks (.csv), everything (.json).

## The metered lanes
Two tasks on gpt-terra, metered per run, routed by `task`. Input string fields: `task` (design or cutover), `platform`, `sandbox`, `carry` (cutover only: the design's decisions) and `facts`.
- `design` replies with `summary`, `isolation`, `simulation`, `catalog` (every outcome with its operation, trigger, kind and code), `events`, `time`, `reset`, `keys`, `quotas`, `abuse`, `parity` and `caveats`.
- `cutover` replies with `summary`, `gate`, `checklist`, `canary`, `pitfalls`, `go_back`, `metrics`, `docs` and `caveats`.
The page holds every reply to its own facts: the ranked rungs, every outcome covered with a trigger, a real code and an operation, no invented card number, one meaning per value, every event fired, time simulated, _test_ and _live_ prefixes; the settled gate, a named pre-approval error, the four checklist steps, a canary confirming the event delivery, and exactly the five pitfalls.

## Links
- API tutorial: https://sandbox-gate.skillsafe.ai/api.html
- Source skill: https://skillsafe.ai/skill/@samber/api-test-mode-design