Write HTML. Render video. Built for agents.
JavaScript aiai-agentsanimationclaude-codeffmpegframeworkgsaphtmlhtml-to-videomcpmotion-graphicspuppeteerrenderingtypescriptvideo-editor
import { failCommand } from "../../utils/commandResult.js";
/**
* Shared CLI error boundary for `hyperframes figma` subcommands: typed
* client errors (NO_TOKEN, BAD_TOKEN, …) and input errors (bad ref, bad
* format) all carry actionable, user-facing messages — present them via
* the CLI's standard errorBox, not a stack trace. Non-Error throws still
* surface raw.
*
* It reports inline to name the typed code (FigmaClientError code) — the whole
* first-run funnel: NO_TOKEN → later success is onboarding conversion.
*/
import { FigmaClientError } from "@hyperframes/core/figma";
import { errorBox } from "../../ui/format.js";
export async function withFigmaErrors(command: string, fn: () => Promise<void>): Promise<void> {
try {
await fn();
} catch (err) {
if (err instanceof Error) {
try {
const { trackCommandFailure } = await import("../../telemetry/events.js");
// Surface the typed code (NO_TOKEN, BAD_TOKEN, RATE_LIMITED, …) as the
// error name — `FigmaClientError` alone says nothing in a dashboard.
trackCommandFailure(
command,
err,
err instanceof FigmaClientError ? { error_name: err.code, endpoint: err.endpoint } : {},
);
} catch {
// Telemetry must never mask the real command failure.
}
const [title = "figma command failed", ...rest] = err.message.split("\n");
errorBox(title, rest.length > 0 ? rest.join("\n") : undefined);
failCommand(1, err);
}
throw err;
}
}