feat: create the initial platform owner from an operator command

A fresh production installation deadlocked. ADR 0034 gave Ekklesia a login
page and ADR 0019 gave it a single-use platform-owner bootstrap, but that
bootstrap grants ownership to a user id that already exists and has no way
to bring one into being. Public signup is disabled by default and
production cannot enable it at all, so a newly deployed instance had a
login page, no account, and no supported way to create one — only hand-
written SQL or briefly opening public registration.

    pnpm identity:create-initial-owner -- --email [email protected]

creates the identity and grants it ownership, in the same shape as the
other operator commands: trusted local tooling, reachable only by someone
who already holds the database credentials, serving no HTTP route. No
/signup, /register, or /setup route is added, no bootstrap endpoint, and
no setup token; public signup stays disabled in every environment and a
test asserts a freshly bootstrapped deployment still refuses
/api/auth/sign-up/email.

The identity is created through Better Auth's own sign-up path, so
hashing, salting, normalization and the 8-128 bound are the identity
system's rather than a second implementation that could drift. Because the
server's instance refuses account creation, the command builds its own
in-process instance; nothing in apps/api imports it and the running
server's posture is untouched.

The sign-up result is verified rather than trusted. Better Auth returns a
fabricated success for an already-registered address — a user object whose
id was never persisted, database unchanged. That is correct enumeration
defence on a public endpoint and a trap here, so the identity is always
read back from the database and that read, not the response, is what the
grant uses.

An existing usable identity is adopted: ownership is granted and its
password left alone, reported as an adoption rather than a creation. This
is the recovery path for the one partial failure possible, since sign-up
commits on Better Auth's connection and the grant is a separate
transaction. A credential-less row is refused instead — making it an owner
would consume the single use on an identity nobody can authenticate as. A
password is never reset and an account is never taken over.

Whole invocations are serialized by an advisory lock on their own key, so
a race cannot leave a stray signable-in account behind; the grant's own
locked single-use transaction is unchanged. There is no --force or reset.

The password is prompted for invisibly or read from stdin, never from
argv (--password is refused with an explanation) and never from an
environment variable, so nothing about it becomes permanent deployment
configuration. The command needs DATABASE_URL and nothing else — verified
that sign-up needs no secret, base URL, or trusted origins, and that the
resulting account signs in through a separately configured server — so the
first owner can be created before TLS or DNS exist. It never migrates.

Proven end to end rather than at the database: a fresh installation is
bootstrapped, signed in through the real browser login transport with
signup disabled, its principal read, and used to open an owner-protected
resource that stays closed to anonymous callers.

Co-Authored-By: Claude Opus 5 <[email protected]>

core/src/authorization/index.ts

@@ -1,6 +1,7 @@
 export * from "./bootstrap.js";
 export * from "./database.js";
 export * from "./evaluator.js";
+export * from "./initial-owner.js";
 export * from "./permissions.js";
 export * from "./ports.js";
 export * from "./store.js";

core/src/authorization/initial-owner.ts

@@ -0,0 +1,278 @@
+import type { EkklesiaUserId } from "@ekklesia/contracts";
+import { safeParseEkklesiaUserId } from "@ekklesia/contracts";
+import { and, eq, sql } from "drizzle-orm";
+import type { BootstrapIdentityCreator } from "../identity/bootstrap-identity.js";
+import { BootstrapIdentityError } from "../identity/bootstrap-identity.js";
+import type { IdentityDrizzleDatabase } from "../identity/database.js";
+import { ekklesiaCoreAccount, ekklesiaCoreUser } from "../identity/schema.js";
+import { bootstrapPlatformOwner, PlatformOwnerBootstrapError } from "./bootstrap.js";
+import type { AuthorizationDrizzleDatabase } from "./database.js";
+import { ekklesiaCorePlatformOwnerBootstrap } from "./schema.js";
+
+export type InitialPlatformOwnerErrorCode =
+  | "INITIAL_OWNER_ALREADY_COMPLETED"
+  | "INITIAL_OWNER_EMAIL_INVALID"
+  | "INITIAL_OWNER_IDENTITY_NOT_CREATED"
+  | "INITIAL_OWNER_IDENTITY_UNUSABLE"
+  | "INITIAL_OWNER_OWNERSHIP_GRANT_FAILED"
+  | "INITIAL_OWNER_PASSWORD_TOO_LONG"
+  | "INITIAL_OWNER_PASSWORD_TOO_SHORT"
+  | "INITIAL_OWNER_SCHEMA_UNAVAILABLE";
+
+const initialOwnerMessages: Readonly<Record<InitialPlatformOwnerErrorCode, string>> = Object.freeze(
+  {
+    INITIAL_OWNER_ALREADY_COMPLETED:
+      "This installation already has a platform owner. Initial-owner bootstrap runs once and cannot run again.",
+    INITIAL_OWNER_EMAIL_INVALID: "The email address is not a valid address.",
+    INITIAL_OWNER_IDENTITY_NOT_CREATED:
+      "The identity system reported success but no account exists for that address. Nothing was changed; the bootstrap is still available.",
+    INITIAL_OWNER_IDENTITY_UNUSABLE:
+      "An account already exists for that address but has no password and could never sign in. Choose a different address, or remove that account before retrying.",
+    INITIAL_OWNER_OWNERSHIP_GRANT_FAILED:
+      "The identity exists but platform ownership was not granted. Bootstrap is still available: run this command again with the same email address to finish it.",
+    INITIAL_OWNER_PASSWORD_TOO_LONG: "The password is longer than the identity system allows.",
+    INITIAL_OWNER_PASSWORD_TOO_SHORT: "The password is shorter than the identity system allows.",
+    INITIAL_OWNER_SCHEMA_UNAVAILABLE:
+      "The database is unreachable or has not been migrated. Apply migrations before creating the initial owner.",
+  },
+);
+
+/**
+ * Never carries the password, a hash, a connection string, SQL, a driver error, or a cause. The
+ * messages are written to be read by an operator at a terminal, so each one states what happened
+ * *and* what is still true about the installation — in particular whether bootstrap remains
+ * available, which is the only fact that determines what they should do next.
+ */
+export class InitialPlatformOwnerError extends Error {
+  readonly code: InitialPlatformOwnerErrorCode;
+
+  constructor(code: InitialPlatformOwnerErrorCode) {
+    super(initialOwnerMessages[code]);
+    this.name = "InitialPlatformOwnerError";
+    this.code = code;
+    this.stack = `${this.name}: ${this.message}`;
+    Object.freeze(this);
+  }
+}
+
+// A CLI-level advisory key distinct from `bootstrap.ts`'s: taking the same one would deadlock this
+// operation against the grant it performs. This serializes whole invocations; that one serializes
+// the grant transaction itself, and both remain in force.
+const initialOwnerLockNamespace = 1_500_000_003;
+const initialOwnerLockKey = 1_500_000_004;
+
+export interface CreateInitialPlatformOwnerDependencies {
+  readonly authorizationDb: AuthorizationDrizzleDatabase;
+  readonly identityDb: IdentityDrizzleDatabase;
+  readonly createIdentity: BootstrapIdentityCreator;
+  /** Injected so a partial failure — identity created, ownership not granted — can be exercised.
+   * Production composition leaves it unset and the real single-use bootstrap is used. */
+  readonly grantOwnership?: (userId: EkklesiaUserId) => Promise<void>;
+}
+
+export interface CreateInitialPlatformOwnerInput {
+  readonly email: string;
+  readonly password: string;
+  /** The display name the identity model requires. */
+  readonly name: string;
+}
+
+export interface InitialPlatformOwnerResult {
+  readonly email: string;
+  readonly userId: EkklesiaUserId;
+  /** `created` when this call made the identity; `adopted` when one already existed for that
+   * address and only ownership was granted. Reported so an operator is never told something was
+   * created when it was not. */
+  readonly identity: "adopted" | "created";
+}
+
+interface ExistingIdentity {
+  readonly userId: string;
+  readonly hasCredential: boolean;
+  /** As the identity system actually stored it, which is not necessarily as the operator typed it. */
+  readonly storedEmail: string;
+}
+
+/**
+ * Finds an identity by email, case-insensitively.
+ *
+ * Deliberately not an exact match on the caller's spelling: Better Auth lowercases the address it
+ * stores (verified against the pinned release — `[email protected]` persists as
+ * `[email protected]`), so an exact comparison would fail to find the very account this operation
+ * had just created and would report it missing.
+ */
+async function findIdentityByEmail(
+  db: IdentityDrizzleDatabase,
+  email: string,
+): Promise<ExistingIdentity | undefined> {
+  const rows = await db
+    .select({ email: ekklesiaCoreUser.email, id: ekklesiaCoreUser.id })
+    .from(ekklesiaCoreUser)
+    .where(sql`lower(${ekklesiaCoreUser.email}) = ${email.trim().toLowerCase()}`)
+    .limit(1);
+  const found = rows[0];
+  if (!found) {
+    return undefined;
+  }
+  // A row in `user` alone is not an account anyone can sign in as — the credential lives in
+  // `account`. The distinction matters: granting ownership to a passwordless row would consume the
+  // one-time bootstrap on an identity nobody can authenticate as, leaving the installation with an
+  // owner that cannot log in and no second chance.
+  const credentials = await db
+    .select({ id: ekklesiaCoreAccount.id })
+    .from(ekklesiaCoreAccount)
+    .where(
+      and(
+        eq(ekklesiaCoreAccount.userId, found.id),
+        eq(ekklesiaCoreAccount.providerId, "credential"),
+      ),
+    )
+    .limit(1);
+  return { hasCredential: credentials.length > 0, storedEmail: found.email, userId: found.id };
+}
+
+function translateIdentityCreationError(error: unknown): InitialPlatformOwnerError {
+  if (error instanceof BootstrapIdentityError) {
+    if (error.code === "BOOTSTRAP_IDENTITY_PASSWORD_TOO_SHORT") {
+      return new InitialPlatformOwnerError("INITIAL_OWNER_PASSWORD_TOO_SHORT");
+    }
+    if (error.code === "BOOTSTRAP_IDENTITY_PASSWORD_TOO_LONG") {
+      return new InitialPlatformOwnerError("INITIAL_OWNER_PASSWORD_TOO_LONG");
+    }
+    if (error.code === "BOOTSTRAP_IDENTITY_EMAIL_INVALID") {
+      return new InitialPlatformOwnerError("INITIAL_OWNER_EMAIL_INVALID");
+    }
+  }
+  return new InitialPlatformOwnerError("INITIAL_OWNER_IDENTITY_NOT_CREATED");
+}
+
+/**
+ * Creates the first platform owner of a fresh installation: one identity, then the single-use
+ * ownership grant.
+ *
+ * ## Why the identity is verified rather than trusted
+ *
+ * Better Auth's sign-up returns a **fabricated success** when the address is already registered —
+ * a user object carrying an id that was never persisted, with the database left unchanged (verified
+ * against the pinned release). On a public endpoint that is correct: it refuses to tell an
+ * anonymous caller which addresses exist. Here it is a trap, because a caller that believed the
+ * response would report an owner it never created and then grant ownership to an id that does not
+ * exist. So the identity is always read back from the database, and that read — not the sign-up
+ * response — is what the grant uses.
+ *
+ * ## Why an existing identity is adopted rather than refused or overwritten
+ *
+ * If the address already has a usable account and bootstrap is still open, ownership is granted to
+ * it and **its password is left completely alone**. This is precisely the recovery path for the one
+ * partial failure this operation can have: sign-up commits through Better Auth's own connection and
+ * the grant runs in a separate transaction, so a crash between them leaves an identity with no
+ * ownership. Refusing would strand a fresh installation with no way forward but manual SQL; a
+ * password reset would be a silent takeover of an account the operator may not own. Adoption
+ * changes no credential, and it is reported as `adopted` so nobody is told an account was created
+ * when it was not.
+ *
+ * That is safe here specifically because production has **no** other way to create an identity —
+ * public signup cannot be enabled at all — so an account present on a fresh installation was put
+ * there by someone who already had database access.
+ *
+ * ## Ordering
+ *
+ * The bootstrap sentinel is checked before an identity is created, so a closed bootstrap never
+ * leaves a stray account behind. That check is an early exit, not the authority: `bootstrapPlatformOwner`
+ * re-checks it inside its own locked transaction, which is what actually makes the grant single-use.
+ */
+export async function createInitialPlatformOwner(
+  dependencies: CreateInitialPlatformOwnerDependencies,
+  input: CreateInitialPlatformOwnerInput,
+): Promise<InitialPlatformOwnerResult> {
+  const email = input.email.trim();
+  if (email.length === 0) {
+    throw new InitialPlatformOwnerError("INITIAL_OWNER_EMAIL_INVALID");
+  }
+
+  // Serializes entire invocations against each other. Without it two operators racing on a fresh
+  // installation would both pass the sentinel check and both create an identity, and only then
+  // would one lose the grant — leaving a stray account that can sign in and owns nothing. The grant
+  // alone was already safe; this makes the whole command safe.
+  return dependencies.authorizationDb.transaction(async (lockTx) => {
+    await lockTx.execute(
+      sql`select pg_advisory_xact_lock(${initialOwnerLockNamespace}, ${initialOwnerLockKey})`,
+    );
+
+    let alreadyConsumed: boolean;
+    try {
+      const consumed = await lockTx
+        .select({ id: ekklesiaCorePlatformOwnerBootstrap.id })
+        .from(ekklesiaCorePlatformOwnerBootstrap)
+        .limit(1);
+      alreadyConsumed = consumed.length > 0;
+    } catch {
+      // An unreachable database and a database without the authorization tables are the same
+      // situation to an operator running this on a fresh server: the deployment step before this
+      // one has not happened. Never migrated implicitly — see ADR 0035.
+      throw new InitialPlatformOwnerError("INITIAL_OWNER_SCHEMA_UNAVAILABLE");
+    }
+    if (alreadyConsumed) {
+      throw new InitialPlatformOwnerError("INITIAL_OWNER_ALREADY_COMPLETED");
+    }
+
+    let existing: ExistingIdentity | undefined;
+    try {
+      existing = await findIdentityByEmail(dependencies.identityDb, email);
+    } catch {
+      throw new InitialPlatformOwnerError("INITIAL_OWNER_SCHEMA_UNAVAILABLE");
+    }
+    if (existing && !existing.hasCredential) {
+      throw new InitialPlatformOwnerError("INITIAL_OWNER_IDENTITY_UNUSABLE");
+    }
+
+    // Captured before the read-back below, which cannot distinguish the two: an identity this call
+    // just created and one it adopted both come back with a credential.
+    const identity: "adopted" | "created" = existing ? "adopted" : "created";
+
+    if (!existing) {
+      try {
+        await dependencies.createIdentity({
+          email,
+          name: input.name,
+          password: input.password,
+        });
+      } catch (error) {
+        throw translateIdentityCreationError(error);
+      }
+      existing = await findIdentityByEmail(dependencies.identityDb, email).catch(() => undefined);
+      if (!existing) {
+        // Reached when sign-up reported success without persisting anything.
+        throw new InitialPlatformOwnerError("INITIAL_OWNER_IDENTITY_NOT_CREATED");
+      }
+    }
+
+    const parsedUserId = safeParseEkklesiaUserId(existing.userId);
+    if (!parsedUserId.success) {
+      throw new InitialPlatformOwnerError("INITIAL_OWNER_IDENTITY_NOT_CREATED");
+    }
+
+    try {
+      const grant =
+        dependencies.grantOwnership ??
+        ((userId: EkklesiaUserId) =>
+          bootstrapPlatformOwner(dependencies.authorizationDb, { userId }));
+      await grant(parsedUserId.data);
+    } catch (error) {
+      if (
+        error instanceof PlatformOwnerBootstrapError &&
+        error.code === "PLATFORM_OWNER_BOOTSTRAP_ALREADY_CONSUMED"
+      ) {
+        throw new InitialPlatformOwnerError("INITIAL_OWNER_ALREADY_COMPLETED");
+      }
+      // The identity is real and remains usable; the sentinel rolled back with the grant, so
+      // re-running this command with the same address adopts it and finishes.
+      throw new InitialPlatformOwnerError("INITIAL_OWNER_OWNERSHIP_GRANT_FAILED");
+    }
+
+    // The stored address, never the operator's spelling: the identity system lowercases what it
+    // persists, and telling someone to sign in as `[email protected]` when the account is
+    // `[email protected]` invites them to conclude the command failed.
+    return Object.freeze({ email: existing.storedEmail, identity, userId: parsedUserId.data });
+  });
+}

core/src/cli/create-initial-owner.ts

@@ -0,0 +1,233 @@
+import path from "node:path";
+import { pathToFileURL } from "node:url";
+import {
+  createAuthorizationDrizzleDatabase,
+  createInitialPlatformOwner,
+  InitialPlatformOwnerError,
+} from "../authorization/index.js";
+import {
+  createBootstrapIdentityCreator,
+  createIdentityDrizzleDatabase,
+} from "../identity/index.js";
+import { resolveExecutableDatabaseTransport } from "./database-transport-environment.js";
+
+/**
+ * Creates the first platform owner of a fresh installation — one identity, then the single-use
+ * ownership grant — so a freshly deployed Ekklesia has a supported way to get its first
+ * administrator (ADR 0035).
+ *
+ * This exists because the login page has nothing to log in to on a new installation. Public signup
+ * is disabled in every environment and production cannot enable it at all, which is correct and is
+ * not changed here; the consequence was simply that no supported path created the first account.
+ * This is that path, and it is deliberately the same shape as `pnpm authorization:bootstrap-owner`
+ * and `pnpm modules:enable`: explicit, operator-invoked, never HTTP-exposed, never inferred, and
+ * reachable only by someone who already holds the database credentials.
+ *
+ * It creates no HTTP route, starts no server, and applies no migration.
+ */
+class InitialOwnerCliError extends Error {
+  constructor(message: string) {
+    super(message);
+    this.name = "InitialOwnerCliError";
+    this.stack = `${this.name}: ${this.message}`;
+  }
+}
+
+const usage =
+  "Usage: pnpm identity:create-initial-owner -- --email <address> [--name <display name>]\n" +
+  "The password is never taken from the command line. It is prompted for when a terminal is\n" +
+  "attached, and otherwise read from standard input:\n" +
+  "  printf '%s' \"$PASSWORD\" | pnpm identity:create-initial-owner -- --email [email protected]";
+
+interface ParsedArgs {
+  readonly email: string;
+  readonly name?: string;
+}
+
+/** A minimal, hand-rolled flag parser, exactly like `bootstrap-platform-owner.ts` and
+ * `set-module-enabled.ts`. There is deliberately no `--password`: see `readPassword`. */
+function parseArgs(argv: readonly string[]): ParsedArgs {
+  let email: string | undefined;
+  let name: string | undefined;
+  for (let index = 0; index < argv.length; index += 1) {
+    const flag = argv[index];
+    if (flag === "--email") {
+      email = argv[index + 1];
+      index += 1;
+    } else if (flag === "--name") {
+      name = argv[index + 1];
+      index += 1;
+    } else if (flag === "--password" || flag?.startsWith("--password=") === true) {
+      // Refused explicitly rather than ignored, so an operator who tries it learns why instead of
+      // wondering why their password was not used.
+      throw new InitialOwnerCliError(
+        "--password is not supported: a command-line password is visible in shell history and in the process list.\n" +
+          usage,
+      );
+    }
+  }
+  if (email === undefined || email.trim().length === 0) {
+    throw new InitialOwnerCliError(usage);
+  }
+  return name === undefined ? { email } : { email, name };
+}
+
+/** Reads one line without echoing it. No dependency is added for this: the command needs exactly
+ * one hidden field, and Node's own raw-mode stdin is enough. */
+function promptHidden(prompt: string): Promise<string> {
+  return new Promise<string>((resolve, reject) => {
+    const input = process.stdin;
+    process.stdout.write(prompt);
+    input.setRawMode(true);
+    input.resume();
+    input.setEncoding("utf8");
+    let value = "";
+    const finish = (settle: () => void): void => {
+      input.removeListener("data", onData);
+      input.setRawMode(false);
+      input.pause();
+      process.stdout.write("\n");
+      settle();
+    };
+    const onData = (chunk: string): void => {
+      for (const character of chunk) {
+        if (character === "\r" || character === "\n" || character === "\u0004") {
+          finish(() => resolve(value));
+          return;
+        }
+        if (character === "\u0003") {
+          finish(() => reject(new InitialOwnerCliError("Cancelled. No changes were made.")));
+          return;
+        }
+        // Backspace and delete, so a mistyped character can be corrected in a field nobody can see.
+        if (character === "\u007f" || character === "\b") {
+          value = value.slice(0, -1);
+          continue;
+        }
+        value += character;
+      }
+    };
+    input.on("data", onData);
+  });
+}
+
+/** Reads standard input to end, for automated deployment. */
+function readStdin(): Promise<string> {
+  return new Promise<string>((resolve, reject) => {
+    let buffer = "";
+    process.stdin.setEncoding("utf8");
+    process.stdin.on("data", (chunk: string) => {
+      buffer += chunk;
+    });
+    process.stdin.on("end", () => resolve(buffer));
+    process.stdin.on("error", () =>
+      reject(new InitialOwnerCliError("Standard input could not be read.")),
+    );
+  });
+}
+
+/**
+ * Obtains the password without it ever appearing in `argv`, an environment variable, or the
+ * terminal.
+ *
+ * A command-line password is visible in shell history and to every other process on the machine via
+ * the process list, and an environment variable tends to become permanent — a bootstrap password
+ * belongs to one invocation, not to the deployment's configuration, so this command deliberately
+ * reads no environment variable for it and none is added to `.env.example`.
+ *
+ * Interactive use asks twice and compares, because the field is invisible and a typo would create
+ * an owner nobody can sign in as, on an installation whose one bootstrap has now been consumed.
+ */
+async function readPassword(): Promise<string> {
+  if (process.stdin.isTTY === true) {
+    const password = await promptHidden("Password: ");
+    const confirmation = await promptHidden("Confirm password: ");
+    if (password !== confirmation) {
+      throw new InitialOwnerCliError("The passwords did not match. No changes were made.");
+    }
+    return password;
+  }
+  // A single trailing newline is stripped so `printf '%s\n'`, a here-string, and an editor-produced
+  // file all behave the same; nothing else is trimmed, because leading and trailing spaces are
+  // legitimate password characters.
+  const piped = await readStdin();
+  return piped.replace(/\r?\n$/u, "");
+}
+
+async function main(): Promise<void> {
+  const { email, name } = parseArgs(process.argv.slice(2));
+
+  // Read only when this command actually executes — never merely by importing this module.
+  const databaseUrl = process.env.DATABASE_URL;
+  if (typeof databaseUrl !== "string" || databaseUrl.trim().length === 0) {
+    throw new InitialOwnerCliError("DATABASE_URL is required.");
+  }
+
+  const password = await readPassword();
+  if (password.length === 0) {
+    throw new InitialOwnerCliError("A password is required. No changes were made.");
+  }
+
+  // Asserted before any connection is constructed, so an unsafe production transport never opens a
+  // client at all — identical to every other Core executable.
+  const transport = resolveExecutableDatabaseTransport(databaseUrl);
+  const databaseOptions = {
+    connectionTimeoutMs: 10_000,
+    operationTimeoutMs: 30_000,
+    transport,
+  };
+  const authorization = createAuthorizationDrizzleDatabase(databaseUrl, databaseOptions);
+  const identity = createIdentityDrizzleDatabase(databaseUrl, databaseOptions);
+  try {
+    const result = await createInitialPlatformOwner(
+      {
+        authorizationDb: authorization.db,
+        createIdentity: createBootstrapIdentityCreator(identity.db),
+        identityDb: identity.db,
+      },
+      { email, name: name ?? email, password },
+    );
+
+    // Never the password, its length, a hash, a session token, a user id, or the database URL.
+    console.log(
+      result.identity === "created"
+        ? "Initial platform owner created successfully."
+        : "An account already existed for that address; platform ownership was granted to it and its password was left unchanged.",
+    );
+    console.log(`Email: ${result.email}`);
+    console.log("You can now sign in to the Ekklesia dashboard.");
+  } finally {
+    await Promise.all([
+      authorization.close().catch(() => undefined),
+      identity.close().catch(() => undefined),
+    ]);
+  }
+}
+
+function reportInitialOwnerCliError(error: unknown): void {
+  if (error instanceof InitialOwnerCliError) {
+    console.error(error.message);
+    return;
+  }
+  if (error instanceof InitialPlatformOwnerError) {
+    console.error(`[${error.code}] ${error.message}`);
+    return;
+  }
+  // Deliberately generic: the caught value may carry SQL, a connection string, a driver detail, or
+  // the password itself if it were ever interpolated into a message upstream.
+  console.error("Initial platform owner creation failed unexpectedly.");
+}
+
+function isDirectExecution(): boolean {
+  const entrypoint = process.argv[1];
+  return (
+    entrypoint !== undefined && import.meta.url === pathToFileURL(path.resolve(entrypoint)).href
+  );
+}
+
+if (isDirectExecution()) {
+  main().catch((error: unknown) => {
+    reportInitialOwnerCliError(error);
+    process.exitCode = 1;
+  });
+}

core/src/identity/bootstrap-identity.ts

@@ -0,0 +1,132 @@
+import { randomBytes } from "node:crypto";
+import { betterAuth } from "better-auth";
+import { drizzleAdapter } from "better-auth/adapters/drizzle";
+import type { IdentityDrizzleDatabase } from "./database.js";
+import { identityAdapterSchema } from "./schema.js";
+
+/**
+ * Creates a first credential identity **through Better Auth's own sign-up path**, for the
+ * operator-invoked initial-owner bootstrap and nothing else.
+ *
+ * ## Why this is not `createIdentityRuntime`
+ *
+ * The server runtime deliberately refuses to create accounts: `disableSignUp` is derived from
+ * `IdentityConfig.publicSignupEnabled`, which production cannot enable at all
+ * (`PUBLIC_SIGNUP_UNSUPPORTED_IN_PRODUCTION`). That is correct and is not relaxed here — calling
+ * `auth.api.signUpEmail` on the server's own instance is refused with
+ * `EMAIL_PASSWORD_SIGN_UP_DISABLED`, which was verified against the pinned release rather than
+ * assumed. So a fresh production installation has no way to create the one identity it needs.
+ *
+ * This is that way, and its safety comes from *where it can be invoked*, not from a weaker policy:
+ * it is an in-process library call reachable only by an operator who already holds the database
+ * credentials, it serves no HTTP route, and it is mounted nowhere. Nothing in `apps/api` imports
+ * it. Public signup remains disabled in every environment.
+ *
+ * ## Why the library path rather than writing rows
+ *
+ * Hashing, salting, email normalization, the `user`/`account` pair, and the 8–128 password bounds
+ * are all Better Auth's, applied exactly as they are for a browser sign-up. Hand-writing a user row
+ * and a password hash would be a second, divergent identity implementation — and the failure mode
+ * is an account that exists but cannot sign in.
+ *
+ * ## Prerequisites this deliberately does not have
+ *
+ * No `baseURL`, no trusted origins, and no configured identity secret. All three were tested
+ * against the pinned release: sign-up succeeds without them, and the resulting account signs in
+ * through a *separately constructed* instance carrying a different secret — password hashes are
+ * salted per account and independent of the secret. That matters operationally, because it means an
+ * operator can create the first owner on a freshly provisioned server before DNS, TLS, or the
+ * public origin exist. The ephemeral secret below exists only so the instance is explicitly
+ * configured rather than silently defaulted; this instance never mints a session, so nothing is
+ * ever signed with it and it is discarded when the process exits.
+ */
+export type BootstrapIdentityErrorCode =
+  | "BOOTSTRAP_IDENTITY_CREATION_FAILED"
+  | "BOOTSTRAP_IDENTITY_EMAIL_INVALID"
+  | "BOOTSTRAP_IDENTITY_PASSWORD_TOO_LONG"
+  | "BOOTSTRAP_IDENTITY_PASSWORD_TOO_SHORT";
+
+const bootstrapIdentityMessages: Readonly<Record<BootstrapIdentityErrorCode, string>> =
+  Object.freeze({
+    BOOTSTRAP_IDENTITY_CREATION_FAILED: "The initial identity could not be created.",
+    BOOTSTRAP_IDENTITY_EMAIL_INVALID: "The email address is not a valid address.",
+    BOOTSTRAP_IDENTITY_PASSWORD_TOO_LONG: "The password is longer than the identity system allows.",
+    BOOTSTRAP_IDENTITY_PASSWORD_TOO_SHORT:
+      "The password is shorter than the identity system allows.",
+  });
+
+/** Never carries the password, the email, a hash, a Better Auth response body, or a cause. */
+export class BootstrapIdentityError extends Error {
+  readonly code: BootstrapIdentityErrorCode;
+
+  constructor(code: BootstrapIdentityErrorCode) {
+    super(bootstrapIdentityMessages[code]);
+    this.name = "BootstrapIdentityError";
+    this.code = code;
+    this.stack = `${this.name}: ${this.message}`;
+    Object.freeze(this);
+  }
+}
+
+export interface BootstrapIdentityInput {
+  readonly email: string;
+  readonly password: string;
+  readonly name: string;
+}
+
+/** Creates one credential identity. Resolving says only that the identity system reported success —
+ * never that a row exists, which the caller must confirm for itself (see `createInitialPlatformOwner`). */
+export type BootstrapIdentityCreator = (input: BootstrapIdentityInput) => Promise<void>;
+
+/** Better Auth reports failures as an `APIError` carrying a `body.code`. Only the codes that are
+ * genuinely the operator's to act on are distinguished; everything else collapses into the generic
+ * creation failure so no library internal, response body, or cause is ever surfaced. */
+function translateBootstrapIdentityError(error: unknown): BootstrapIdentityError {
+  const code = (error as { body?: { code?: unknown } } | null)?.body?.code;
+  if (code === "PASSWORD_TOO_SHORT") {
+    return new BootstrapIdentityError("BOOTSTRAP_IDENTITY_PASSWORD_TOO_SHORT");
+  }
+  if (code === "PASSWORD_TOO_LONG") {
+    return new BootstrapIdentityError("BOOTSTRAP_IDENTITY_PASSWORD_TOO_LONG");
+  }
+  if (code === "VALIDATION_ERROR") {
+    return new BootstrapIdentityError("BOOTSTRAP_IDENTITY_EMAIL_INVALID");
+  }
+  return new BootstrapIdentityError("BOOTSTRAP_IDENTITY_CREATION_FAILED");
+}
+
+export function createBootstrapIdentityCreator(
+  db: IdentityDrizzleDatabase,
+): BootstrapIdentityCreator {
+  const auth = betterAuth({
+    database: drizzleAdapter(db, { provider: "pg", schema: identityAdapterSchema }),
+    emailAndPassword: {
+      // No session is minted here. The operator signs in through the ordinary browser transport
+      // afterwards, which is also what proves the account really works.
+      autoSignIn: false,
+      // Enabled on *this* instance only. See the note above: this object serves no HTTP route.
+      disableSignUp: false,
+      enabled: true,
+      // The identical bounds `createIdentityRuntime` pins, so the CLI can never accept a password
+      // the browser would later reject, or reject one it would accept.
+      maxPasswordLength: 128,
+      minPasswordLength: 8,
+    },
+    logger: {
+      // Better Auth's own log call sites interpolate raw input and exceptions; during account
+      // creation that input includes a password. Nothing it wants to log is ever emitted.
+      disabled: true,
+    },
+    secret: randomBytes(32).toString("base64url"),
+  });
+
+  return async (input: BootstrapIdentityInput): Promise<void> => {
+    try {
+      await auth.api.signUpEmail({
+        body: { email: input.email, name: input.name, password: input.password },
+      });
+    } catch (error) {
+      throw translateBootstrapIdentityError(error);
+    }
+  };
+}

core/src/identity/index.ts

@@ -1,3 +1,4 @@
+export * from "./bootstrap-identity.js";
 export * from "./client-address.js";
 export * from "./config.js";
 export * from "./database.js";

core/tests/cli-create-initial-owner.test.ts

@@ -0,0 +1,40 @@
+import { describe, expect, it, vi } from "vitest";
+
+const poolConstructor = vi.hoisted(() => vi.fn());
+
+vi.mock("pg", () => ({ Pool: poolConstructor }));
+
+describe("create-initial-owner CLI import safety", () => {
+  const originalEnv = { ...process.env };
+
+  it("does not read DATABASE_URL, connect to PostgreSQL, or create anything merely by importing the module", async () => {
+    delete process.env.DATABASE_URL;
+
+    // Under Vitest, process.argv[1] is never this file's own path, so isDirectExecution() is false
+    // exactly as it is for `bootstrap-platform-owner.ts` — no argv manipulation is needed to prove
+    // the guard holds. Importing the command must never prompt, connect, or write.
+    await expect(import("../src/cli/create-initial-owner.js")).resolves.toBeDefined();
+
+    expect(poolConstructor).not.toHaveBeenCalled();
+
+    process.env = originalEnv;
+  });
+
+  it("reads no environment variable for the password", async () => {
+    // The password is prompted for or read from standard input; it is deliberately never taken from
+    // the environment, so nothing can make it permanent deployment configuration. Asserted against
+    // the source rather than behaviour, because the absence of a lookup is what matters.
+    const source = await import("node:fs/promises").then((fs) =>
+      fs.readFile(new URL("../src/cli/create-initial-owner.ts", import.meta.url), "utf8"),
+    );
+
+    const environmentReads = [...source.matchAll(/process\.env\.([A-Z_]+)/gu)].map(
+      (match) => match[1],
+    );
+    expect(environmentReads).toEqual(["DATABASE_URL"]);
+    // No dynamic lookup either, which a named-property scan alone would miss.
+    expect(source).not.toMatch(/process\.env\[/u);
+    // And a password supplied on the command line is refused rather than quietly honoured.
+    expect(source).toContain("--password is not supported");
+  });
+});