Hint 1: Approval is not a free-form sentence
Compare the trusted record of the exact action with the action about to execute.
Hint 1 unlocks after your first failed run.
Hint 2: User access and confirmation are separate
Even a matching approval cannot grant a report the user cannot read.
Hint 2 unlocks after 2 failed runs.
Reveal the reference patch and explanation
export function allowToolCall(call, session, approval) {
return Boolean(session && approval?.active === true
&& call.tool === 'export_report' && call.tenant === session.tenant
&& session.reports.includes(call.report)
&& session.allowedDestinations.includes(call.destination)
&& ['tool','tenant','report','destination'].every(k => call[k] === approval[k]));
}The exact approved r1 export to internal://finance passes. Changing destination fails authorization and approval binding. Changing r1 to r2 fails approval binding even when the session can read both reports. Removing r1 from current session access fails despite matching approval. active=false or a model-supplied approved:true cannot grant authority. Repeating this pure predicate with the same still-active inputs would return true again, so single-use execution requires the separate durable executor boundary.
Compare it with your own fix, then rerun the tests. The editor is not changed when you open this explanation.
The reference patch unlocks after you run code you have edited.
Inspect the actual test harness
The browser and the lab download run the same cases, including cases that catch common wrong fixes, such as comparing with one fixture value instead of the real rule. The download includes the complete test adapter and cryptographic fixtures.
Download cases.mjsHow the browser sandbox runs your code
Your code runs in QuickJS 2025-09-13 (quickjs-emscripten 0.32.0) inside a Web Worker, isolated from the page, with a 32 MiB memory limit and a 2.5-second time limit. The test cases and the pass or fail verdict run outside that sandbox, so changing built-ins such as Array.prototype.push cannot change the result. The lab download runs the same cases under Node.js, where your code runs directly.
The same as in current Node.js:
- ES2025 syntax and built-ins, including Set methods such as intersection, iterator helpers, Promise.try, RegExp.escape and the RegExp v flag
Added to the sandbox so that it behaves like Node.js:
- structuredClone, Array.fromAsync, queueMicrotask, atob and btoa
- URL and URLSearchParams, parsed by your browser’s own URL implementation
- TextEncoder and TextDecoder for UTF-8
- Intl.Collator and a locale-aware String.prototype.localeCompare
- console.log and related methods, whose output appears with your results
Not available in the browser sandbox:
- Timers (setTimeout, setInterval), fetch and other network or storage APIs; the exercises are deterministic and offline
- Node.js modules and globals: import, require, process and Buffer
- Web Crypto (crypto.getRandomValues, crypto.subtle); each workshop supplies the cryptographic operations it needs
- Intl other than Collator: toLocaleString and toLocaleDateString ignore the locale, so numbers have no digit grouping and dates use one fixed format in the device’s time zone
- using declarations, Symbol.dispose and DisposableStack (explicit resource management)
- Uint8Array.fromBase64 and toBase64, JSON.rawJSON, Atomics and Error.captureStackTrace
This is the authorization predicate; no export is sent. Production must validate tool schemas, canonicalize approved arguments, protect the approval store, atomically consume approval, prevent replay and reauthorize at execution time. Human approval does not replace authorization.
Your next task: Can a retry, delegated agent or changed argument reuse an approval in your tool system?