/**
 * @file
 *
 * The distinct error thrown when the harness-owned Obsidian instance is gone —
 * it exited, and the CDP endpoint it was serving answers nothing.
 *
 * It exists to replace an anonymous cascade with one named cause. Before it,
 * every test file scheduled after a mid-run death failed in ~60–90 ms with
 * `TypeError: fetch failed` wrapping `connect ECONNREFUSED 127.0.0.1:<port>` —
 * measured at **156 such failures across 21 files** in `obsidian-patterns` on
 * 2026-09-05, not one of them a test result. The absence of an app is not a test
 * outcome, and this error says so in the words a reader needs: the instance
 * died, here is its exit code, and these failures are not about your code.
 *
 * Exported so callers can `instanceof`-match it, like `SilentAsarFallbackError`
 * and `RendererFailedToInitializeError`.
 */
import type { OwnedInstanceExitMarker } from './owned-instance-exit-marker.cjs';
/**
 * Parameters for the {@link OwnedInstanceExitedError} constructor.
 *
 * Everything but the URL is optional: the exit details come from the process
 * that launched the instance, and a caller in another process has them only
 * when the exit marker survived (see `owned-instance-exit-marker.ts`).
 */
export interface OwnedInstanceExitedErrorConstructorParams {
    /**
    The CDP URL the dead instance was serving.
     */
    readonly cdpUrl: string;
    /**
    Exit code, when known.
     */
    readonly code?: null | number | undefined;
    /**
    When the instance exited (`Date.now()` epoch milliseconds), when known.
     */
    readonly exitedAtInMilliseconds?: number | undefined;
    /**
    The tail of what the instance wrote to stdout/stderr, when any was captured.
     */
    readonly outputTail?: string | undefined;
    /**
    PID of the dead instance, when known.
     */
    readonly pid?: number | undefined;
    /**
    Terminating signal, when known.
     */
    readonly signal?: NodeJS.Signals | null | undefined;
    /**
    The spawn-failure message when the instance never started, otherwise absent.
     */
    readonly spawnError?: string | undefined;
}
/**
 * Thrown when the harness-owned Obsidian instance has exited and its CDP
 * endpoint is therefore unreachable. Carries whatever is known about the death
 * — exit code, signal, when it happened, and the output tail — so the failure
 * names its own cause instead of a refused connection.
 */
export declare class OwnedInstanceExitedError extends Error {
    /**
    The CDP URL the dead instance was serving.
     */
    readonly cdpUrl: string;
    /**
    Exit code, when known.
     */
    readonly code: null | number | undefined;
    /**
    When the instance exited (`Date.now()` epoch milliseconds), when known.
     */
    readonly exitedAtInMilliseconds: number | undefined;
    /**
    Terminating signal, when known.
     */
    readonly signal: NodeJS.Signals | null | undefined;
    /**
     * Creates the error from whatever is known about the instance's death.
     *
     * @param params - The CDP URL, and the exit details when they are available.
     */
    constructor(params: OwnedInstanceExitedErrorConstructorParams);
}
/**
 * Builds the error from an exit marker, for a process that did not launch the
 * instance and so has nothing else to go on.
 *
 * A missing marker is expected rather than exceptional — it is cleared on every
 * deliberate kill, and a hard enough kill records nothing — so the error is
 * still built, naming the instance without claiming to know how it died.
 *
 * @param cdpUrl - The CDP URL the dead instance was serving.
 * @param marker - The exit marker read for its port, or `undefined` when there is none.
 * @returns The error to throw in place of a refused connection.
 */
export declare function buildOwnedInstanceExitedErrorFromMarker(cdpUrl: string, marker: OwnedInstanceExitMarker | undefined): OwnedInstanceExitedError;