/**
 * The shared post-install step every harness's `install` command runs.
 *
 * Every install wires the same **developer-experience** surface: the Ory
 * skills and commands, the local Ory stack, the MCP server, and local activity logging.
 * None of that needs an Ory project, so none of it is optional or configurable
 * here — it is simply what installing the plugin gives you.
 *
 * On top of that sits **Ory Agent Security** — user sign-in, permission checks
 * on every tool call, and delegation recording. It runs only when the install
 * has an Ory project URL. The broker and public OAuth2 client id default to the
 * production Agent Security origin and reserved login client; explicit flags
 * select staging, local, or custom deployments. Pass the project URL to
 * `install` (or `configure` later) and the next session signs in and starts
 * checking; leave it out and the
 * plugin logs activity locally and gates nothing. Nothing else changes either way.
 *
 * Everything *else* an Ory project needs to serve Agent Security — the login
 * OAuth2 client, the permission model, native-tool blocks, MCP grants,
 * and the observe/enforce posture — is provisioned in the **Ory Console**
 * (Agent Security), by someone with access to the project. The plugin never
 * writes project configuration, which is why installing it needs only a
 * project URL and no workspace privilege at all.
 */
/**
 * Apply the install's connection intent and print the closing summary.
 *
 * With no connection flags the plugin installs with Agent Security simply not
 * connected. With any of them it delegates to {@link runConfigureCommand},
 * which validates the project URL and any explicit broker override exactly as
 * a bare `configure` would.
 */
export declare function runPostInstall(binName: string, _harness: string, args?: string[]): Promise<void>;