Persist gateway secrets and worktree state in user data

Dockerfile

@@ -10,11 +10,13 @@ RUN dnf -y install ca-certificates git openssh-clients python3 tmux util-linux \
     && dnf clean all
 
 COPY resources/scripts/napseer_mcp_server.py /opt/napseer/napseer_mcp_server.py
+COPY resources/scripts/napseer_credentials.py /opt/napseer/napseer_credentials.py
 COPY resources/scripts/napseer_spake2.py /opt/napseer/napseer_spake2.py
 COPY resources/scripts/terminal/ /opt/napseer/terminal/
 
 RUN python3 -m py_compile \
         /opt/napseer/napseer_mcp_server.py \
+        /opt/napseer/napseer_credentials.py \
         /opt/napseer/napseer_spake2.py \
         /opt/napseer/terminal/__init__.py \
         /opt/napseer/terminal/protocol.py \
@@ -27,6 +29,7 @@ ENV NAPSEER_GATEWAY_PORT=0
 ENV NAPSEER_GATEWAY_PTY_TERMINAL=1
 ENV NAPSEER_SERVICE_ACTIVATION_TIMEOUT_SECONDS=900
 
-VOLUME ["/workspace/.napseer"]
+ENV NAPSEER_USER_DATA_DIR=/var/lib/napseer
+VOLUME ["/var/lib/napseer"]
 
 CMD ["python3", "/opt/napseer/napseer_mcp_server.py", "gateway", "service", "run"]

README.md

@@ -20,3 +20,22 @@ Current state:
   image. It performs no live Napseer API download during the build.
 
 The backend remains the protocol and compatibility source of truth through `/v1/gateway-protocol`.
+
+Private gateway state is stored outside the checkout using the same
+`WorkspacePaths` contract as the CLI. The image sets
+`NAPSEER_USER_DATA_DIR=/var/lib/napseer`; persist that directory in a dedicated
+volume and mount source code separately at `/workspace`. Vaults, worker grants,
+logs, indexes, and process state belong to a worktree; repository signing keys
+are shared across related Git worktrees.
+
+To migrate an existing gateway, stop its old process and retain the old volume.
+Mount that volume at `/workspace/.napseer` and the new persistent volume at
+`/var/lib/napseer`, then run the new image with
+`python3 /opt/napseer/napseer_mcp_server.py auth migrate` as its command.
+Start the service after migration succeeds. Conflicting state is preserved for
+explicit reconciliation. Do not run old and new gateway versions against the
+same credentials during migration.
+
+`resources/scripts/napseer_credentials.py` vendors the CLI's private-state
+contract so the gateway image requires no runtime download. Keep its path and
+migration behavior aligned with the CLI when changing storage layout.

resources/scripts/napseer_credentials.py

@@ -0,0 +1,433 @@
+"""User-owned credentials, separate from repository identity and gateway state.
+
+No network, enrollment, or credential output belongs in this module.
+"""
+from contextlib import contextmanager
+import hashlib
+import json
+import os
+from pathlib import Path
+import stat
+import subprocess
+import sys
+import tempfile
+import threading
+import time
+
+
+class CredentialError(RuntimeError):
+    pass
+
+
+CREDENTIAL_KEYS = {
+    "base_url", "account_id", "token", "access_token", "refresh_token",
+    "token_expires_at", "refresh_expires_at", "account_mode", "oauth_client_id",
+    "oauth_scope", "logged_out", "credential_kind",
+}
+SELECTION_KEYS = {
+    "project_id", "project_slug", "project_name", "project_encryption_state",
+    "project_signing_key_fingerprint", "credential_profile",
+}
+_registry_lock = threading.Lock()
+_thread_locks = {}
+_held = threading.local()
+
+
+def user_data_dir():
+    if os.environ.get("NAPSEER_USER_DATA_DIR"):
+        path = Path(os.environ["NAPSEER_USER_DATA_DIR"]).expanduser()
+        if not path.is_absolute():
+            raise CredentialError("NAPSEER_USER_DATA_DIR must be an absolute path outside the checkout.")
+        return path.resolve()
+    if sys.platform == "win32":
+        return Path(os.environ.get("LOCALAPPDATA", Path.home() / "AppData" / "Local")) / "napseer"
+    if sys.platform == "darwin":
+        return Path.home() / "Library" / "Application Support" / "napseer"
+    xdg = Path(os.environ.get("XDG_DATA_HOME", Path.home() / ".local" / "share"))
+    return (xdg if xdg.is_absolute() else Path.home() / ".local" / "share") / "napseer"
+
+
+class WorkspacePaths:
+    """Pure path resolution: Git identity survives worktree moves, not copies.
+
+    The public locator stays in Git. Repository credentials/signing keys are
+    shared by related worktrees; daemon/vault/cache state belongs to one worktree.
+    No credentials are read and no directory is created during path discovery.
+    """
+    def __init__(self, root=None):
+        start = Path(root or os.environ.get("NAPSEER_PROJECT_ROOT") or Path.cwd()).expanduser().resolve()
+        self.root = start
+        git_dir = common = None
+        environment = {k:v for k,v in os.environ.items() if k not in
+                       {"GIT_DIR", "GIT_WORK_TREE", "GIT_COMMON_DIR", "GIT_INDEX_FILE"}}
+        try:
+            result = subprocess.run(["git", "-C", str(start), "rev-parse", "--path-format=absolute",
+                                     "--show-toplevel", "--git-dir", "--git-common-dir"],
+                                    capture_output=True, text=True, timeout=3, env=environment)
+            if result.returncode == 0:
+                top, git_dir, common = result.stdout.strip().splitlines()
+                self.root = Path(top).resolve()
+        except (OSError, ValueError, subprocess.TimeoutExpired):
+            pass
+        # A nested project's explicit locator is a separate namespace within Git.
+        for candidate in (start, *start.parents):
+            if (candidate / ".napseer/project.json").is_file():
+                self.root = candidate
+                break
+            if git_dir and candidate == self.root:
+                break
+        suffix = str(self.root.relative_to(Path(top).resolve())) if git_dir else ""
+        repo_identity = str(Path(common).resolve()) + ":" + suffix if common else str(self.root)
+        worktree_identity = str(Path(git_dir).resolve()) + ":" + suffix if git_dir else str(self.root)
+        self.repository_id = hashlib.sha256(repo_identity.encode()).hexdigest()
+        self.workspace_id = hashlib.sha256(worktree_identity.encode()).hexdigest()
+        self.data = user_data_dir().expanduser().resolve()
+        if self.data.is_relative_to(self.root) and self.root != Path.home().resolve():
+            raise CredentialError("Napseer user data must be outside the project checkout.")
+        self.legacy = self.root / ".napseer"
+        self.locator = self.legacy / "project.json"
+        self.repository = self.data / "repositories" / self.repository_id
+        self.state = self.data / "workspaces" / self.workspace_id
+        self.signing_key = self.repository / "id_ed25519"
+        self.project_credentials = self.data / "credentials/projects" / (self.repository_id + ".json")
+        old_id = hashlib.sha256(str(self.root).encode()).hexdigest()
+        self.old_credentials = self.data / "credentials/projects" / (old_id + ".json")
+
+    def migration_sources(self):
+        if self.legacy.is_symlink():
+            raise CredentialError("Legacy Napseer state must not be a symbolic link.")
+        sources = []
+        if self.legacy.exists():
+            sources = [p for p in self.legacy.iterdir() if p.name not in {"project.json", ".gitignore"}]
+        if self.old_credentials != self.project_credentials and self.old_credentials.exists():
+            sources.append(self.old_credentials)
+        return sources
+
+    def require_migrated(self):
+        if self.migration_sources():
+            raise CredentialError("Repository-local Napseer state needs migration. Stop older MCP clients and local gateways, then run `nap auth migrate`.")
+
+    def migrate(self, *, gateway=False):
+        """Copy/verify/publish before removing originals; never overwrite conflicts.
+
+        Older processes must be stopped before this explicit operation. Existing
+        daemon PIDs are checked. A partial copy can be retried safely; all source
+        bytes remain until every destination has been verified.
+        """
+        with credential_lock(self.data / "migration"):
+            sources = self.migration_sources()
+            if not sources:
+                return {"status":"already_migrated", "state_dir":str(self.state)}
+            for name in ("gateway.json", "ui.json"):
+                state = read_json(self.legacy / name)
+                pid = state.get("pid")
+                if isinstance(pid, int) and pid > 0:
+                    try:
+                        os.kill(pid, 0)
+                    except ProcessLookupError:
+                        continue
+                    except OSError:
+                        pass
+                    raise CredentialError("Stop the running local gateway/UI before `nap auth migrate`.")
+            entries = []
+            for source in sources:
+                if source.is_symlink() or not source.is_file():
+                    raise CredentialError("Migration requires regular files; original state was preserved.")
+                destination = self.state / source.name
+                if source.name in {"id_ed25519", "id_ed25519.pub"}:
+                    destination = self.repository / source.name
+                if source.name == "auth.json" or source == self.old_credentials:
+                    destination = (self.state / "auth.json") if gateway and source.name == "auth.json" else self.project_credentials
+                original = file_fingerprint(source)
+                copied = None
+                if source.name in {"auth.json", "gateway-auth.json", "state.json"} or source == self.old_credentials:
+                    data = read_json(source)
+                    for field, old, new in (
+                        ("ssh_key_path", self.legacy / "id_ed25519", self.signing_key),
+                        ("vault_path", self.legacy / "vault.json", self.state / "vault.json"),
+                    ):
+                        value = data.get(field)
+                        if isinstance(value, str):
+                            candidate = Path(value).expanduser()
+                            candidate = candidate if candidate.is_absolute() else self.root / candidate
+                            if candidate.resolve() == old.resolve(): data[field] = str(new)
+                    copied = (json.dumps(data, indent=2) + "\n").encode()
+                if destination.is_symlink():
+                    raise CredentialError("Migration destination must not be a symbolic link.")
+                expected = (len(copied), hashlib.sha256(copied).hexdigest()) if copied is not None else original
+                if destination.exists() and file_fingerprint(destination) != expected:
+                    raise CredentialError("Migration conflicts with existing private state; nothing was overwritten. Reconcile the two states before retrying.")
+                if any(dst == destination and digest != expected for _,dst,_,_,digest in entries):
+                    raise CredentialError("Legacy credentials conflict; original state was preserved.")
+                entries.append((source, destination, original, copied, expected))
+            # Publish credentials last, after their keys and other state exist.
+            entries.sort(key=lambda e: e[1] in {self.project_credentials, self.state / "auth.json"})
+            for source, destination, original, copied, expected in entries:
+                private_directory(destination.parent, self.data)
+                if not destination.exists():
+                    fd, temporary = tempfile.mkstemp(prefix=".migrate-", dir=destination.parent)
+                    try:
+                        with os.fdopen(fd, "wb") as stream:
+                            os.fchmod(stream.fileno(), 0o600)
+                            if copied is not None:
+                                stream.write(copied)
+                            else:
+                                with source.open("rb") as original_stream:
+                                    while chunk := original_stream.read(1024 * 1024):
+                                        stream.write(chunk)
+                            stream.flush(); os.fsync(stream.fileno())
+                        # Publish without replacing another login or migration's file.
+                        try:
+                            os.link(temporary, destination)
+                        except FileExistsError:
+                            if file_fingerprint(destination) != expected:
+                                raise CredentialError("Migration destination changed; original state was preserved.") from None
+                    finally:
+                        if os.path.exists(temporary): os.unlink(temporary)
+                if file_fingerprint(destination) != expected or file_fingerprint(source) != original:
+                    raise CredentialError("State changed during migration; original files were preserved. Stop older clients and retry.")
+            for source, destination, original, copied, expected in entries:
+                if file_fingerprint(source) != original or file_fingerprint(destination) != expected:
+                    raise CredentialError("Migration verification failed; remaining originals were preserved.")
+            # Make published directory entries durable before removing originals.
+            # Windows does not support opening directories for fsync this way.
+            if os.name != "nt":
+                parents = set()
+                for _, destination, _, _, _ in entries:
+                    for parent in (destination.parent, *destination.parent.parents):
+                        parents.add(parent)
+                        if parent == self.data: break
+                for parent in sorted(parents, key=lambda p: len(p.parts), reverse=True):
+                    fd = os.open(parent, os.O_RDONLY | getattr(os, "O_DIRECTORY", 0))
+                    try: os.fsync(fd)
+                    finally: os.close(fd)
+            for source, _, _, _, _ in entries:
+                source.unlink()
+            return {"status":"migrated", "state_dir":str(self.state), "files":len(entries)}
+
+
+def private_directory(path, boundary):
+    path, boundary = Path(path), Path(boundary)
+    if not path.is_relative_to(boundary):
+        raise CredentialError("Private state directory is outside its user data boundary.")
+    for parent in (path, *path.parents):
+        if parent.is_symlink():
+            raise CredentialError("Private Napseer state must not use symlink directories.")
+        if parent == boundary: break
+    path.mkdir(mode=0o700, parents=True, exist_ok=True)
+    for parent in (path, *path.parents):
+        parent.chmod(0o700)
+        if parent == boundary: break
+
+
+def file_fingerprint(path):
+    digest = hashlib.sha256()
+    fd = os.open(path, os.O_RDONLY | getattr(os, "O_NOFOLLOW", 0))
+    with os.fdopen(fd, "rb") as stream:
+        if not stat.S_ISREG(os.fstat(stream.fileno()).st_mode):
+            raise CredentialError("Migration requires regular files.")
+        size = 0
+        while chunk := stream.read(1024 * 1024):
+            digest.update(chunk); size += len(chunk)
+    return size, digest.hexdigest()
+
+
+def read_json(path):
+    try:
+        fd = os.open(path, os.O_RDONLY | getattr(os, "O_NOFOLLOW", 0))
+    except FileNotFoundError:
+        return {}
+    except OSError:
+        raise CredentialError("Credential state cannot be opened safely.") from None
+    try:
+        if not stat.S_ISREG(os.fstat(fd).st_mode):
+            raise CredentialError("Credential state must be a regular file.")
+        with os.fdopen(fd, "r", encoding="utf-8") as stream:
+            fd = None
+            raw = stream.read(65537)
+        if len(raw) > 65536:
+            raise CredentialError("Credential state exceeds the size limit.")
+        result = json.loads(raw)
+        if not isinstance(result, dict):
+            raise ValueError()
+        return result
+    except (ValueError, UnicodeError, OSError):
+        raise CredentialError("Credential state is unreadable; run nap auth login to repair it.") from None
+    finally:
+        if fd is not None:
+            os.close(fd)
+
+
+def atomic_json(path, data):
+    path = Path(path)
+    path.parent.mkdir(mode=0o700, parents=True, exist_ok=True)
+    if path.is_symlink():
+        raise CredentialError("Credential state must not be a symbolic link.")
+    fd, temporary = tempfile.mkstemp(prefix=f".{path.name}.", dir=path.parent)
+    try:
+        with os.fdopen(fd, "w", encoding="utf-8") as stream:
+            os.fchmod(stream.fileno(), 0o600)
+            json.dump(data, stream, indent=2)
+            stream.write("\n")
+            stream.flush()
+            os.fsync(stream.fileno())
+        os.replace(temporary, path)
+    finally:
+        if os.path.exists(temporary):
+            os.unlink(temporary)
+
+
+@contextmanager
+def credential_lock(path, timeout=30):
+    """Bounded process lock; nested calls in the same thread reuse the lock."""
+    path = Path(path).absolute()
+    with _registry_lock:
+        thread_lock = _thread_locks.setdefault(str(path), threading.RLock())
+    if not thread_lock.acquire(timeout=timeout):
+        raise CredentialError("Credential update is busy; retry after the other login or refresh finishes.")
+    try:
+        held = getattr(_held, "paths", {})
+        if str(path) in held:
+            yield
+            return
+        path.parent.mkdir(mode=0o700, parents=True, exist_ok=True)
+        try:
+            fd = os.open(str(path) + ".lock", os.O_RDWR | os.O_CREAT | getattr(os, "O_NOFOLLOW", 0), 0o600)
+        except OSError:
+            raise CredentialError("Credential lock cannot be opened safely.") from None
+        try:
+            if not stat.S_ISREG(os.fstat(fd).st_mode):
+                raise CredentialError("Credential lock must be a regular file.")
+            deadline = time.monotonic() + timeout
+            while True:
+                try:
+                    if os.name == "nt":
+                        import msvcrt
+                        os.lseek(fd, 0, os.SEEK_SET)
+                        msvcrt.locking(fd, msvcrt.LK_NBLCK, 1)
+                    else:
+                        import fcntl
+                        fcntl.flock(fd, fcntl.LOCK_EX | fcntl.LOCK_NB)
+                    break
+                except (BlockingIOError, OSError):
+                    if time.monotonic() >= deadline:
+                        raise CredentialError("Credential update is busy; retry after the other login or refresh finishes.") from None
+                    time.sleep(0.025)
+            held[str(path)] = fd
+            _held.paths = held
+            try:
+                yield
+            finally:
+                held.pop(str(path), None)
+        finally:
+            os.close(fd)
+    finally:
+        thread_lock.release()
+
+
+class CredentialStore:
+    def __init__(self, state_dir, legacy_path=None):
+        self.workspace = WorkspacePaths(Path(state_dir).parent)
+        self.state_dir = self.workspace.state
+        self.legacy = Path(legacy_path or self.workspace.legacy / "auth.json").expanduser().absolute()
+        self.user = user_data_dir() / "credentials" / "default.json"
+        self.project = self.workspace.project_credentials
+        self.selection = self.state_dir / "state.json"
+
+    def selected(self):
+        bound = getattr(_held,"bindings",{}).get(str(self.state_dir.absolute()))
+        if bound: return bound
+        explicit = os.environ.get("NAPSEER_AUTH_FILE")
+        if explicit:
+            path = Path(explicit).expanduser().absolute()
+            if path.is_symlink():
+                raise CredentialError("Explicit credentials must not be a symbolic link.")
+            if path.resolve().is_relative_to(self.workspace.root) and self.workspace.root != Path.home().resolve():
+                raise CredentialError("NAPSEER_AUTH_FILE must be outside the checkout; migrate private state first.")
+            return path, "explicit"
+        if self.legacy != self.workspace.legacy / "auth.json":
+            if self.legacy.resolve().is_relative_to(self.workspace.root) and self.workspace.root != Path.home().resolve():
+                raise CredentialError("Explicit credentials must be outside the checkout; migrate private state first.")
+            return self.legacy, "explicit"
+        self.workspace.require_migrated()
+        if read_json(self.selection).get("credential_profile") == "user":
+            return self.user, "user"
+        if self.project.exists() or self.project.is_symlink():
+            return self.project, "project"
+        return self.user, "user"
+
+    @contextmanager
+    def transaction(self):
+        """Pin one credential source through read/refresh/write despite new overrides."""
+        identity = str(self.state_dir.absolute())
+        chosen = self.selected()
+        with credential_lock(chosen[0]):
+            bindings = getattr(_held,"bindings",{})
+            previous = bindings.get(identity)
+            bindings[identity] = chosen; _held.bindings = bindings
+            try: yield
+            finally:
+                if previous: bindings[identity] = previous
+                else: bindings.pop(identity,None)
+
+    def use_user(self):
+        current = read_json(self.user)
+        if not current.get("token") or current.get("logged_out"):
+            raise CredentialError("Run nap auth login before selecting the general user login.")
+        with credential_lock(self.selection):
+            selection = read_json(self.selection)
+            selection["credential_profile"] = "user"
+            atomic_json(self.selection, selection)
+        return self.user
+
+    def migrate(self):
+        self.workspace.migrate()
+        return self.project
+
+    def read(self):
+        path, source = self.selected()
+        credentials = read_json(path)
+        selection = {k: v for k, v in read_json(self.selection).items() if k in SELECTION_KEYS}
+        # Project-bound grants retain their issued project. General grants only
+        # consume repository selection; selection cannot confer authority.
+        return {**selection, **credentials}
+
+    def write(self, data):
+        path, source = self.selected()
+        with credential_lock(path):
+            if source == "explicit":
+                atomic_json(path, data)
+                return
+            credentials = {k: v for k, v in data.items() if k in CREDENTIAL_KEYS}
+            if source == "project":
+                # Preserve project-bound worker/recovery/key metadata during
+                # refresh after legacy migration. Profile selection stays local
+                # to the worktree and must not change sibling worktrees.
+                credentials = {k: v for k, v in data.items() if k != "credential_profile"}
+            current = read_json(path)
+            # Reading project metadata must not churn or overwrite shared
+            # credentials, including after another process logged out.
+            if credentials != current:
+                atomic_json(path, credentials)
+            selection = {**read_json(self.selection), **{k: v for k, v in data.items() if k in SELECTION_KEYS}}
+            if selection != read_json(self.selection):
+                atomic_json(self.selection, selection)
+
+    def login(self, credentials, project=False):
+        target = self.project if project else self.user
+        with credential_lock(target):
+            atomic_json(target, credentials)
+        if project:
+            with credential_lock(self.selection):
+                selection = read_json(self.selection); selection["credential_profile"] = "project"
+                atomic_json(self.selection, selection)
+        return target
+
+    def logout(self, project=False):
+        target = self.selected()[0] if project else self.user
+        with credential_lock(target):
+            previous = read_json(target)
+            # Keep a tombstone: removing an override would silently restore the
+            # more privileged user session on the next request.
+            atomic_json(target, {"base_url": previous.get("base_url"), "logged_out": True})
+        return target

resources/scripts/napseer_mcp_server.py

@@ -49,6 +49,8 @@ import webbrowser
 from datetime import datetime, timedelta, timezone
 from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
 
+from napseer_credentials import WorkspacePaths, credential_lock
+
 from napseer_spake2 import (
     relay_aad,
     relay_context_hash,
@@ -71,31 +73,17 @@ class RelaySocketClosed(RuntimeError):
 
 
 def local_state_dir():
-    preferred = pathlib.Path.cwd() / ".napseer"
-    legacy = pathlib.Path.cwd() / "napseer"
-    if preferred.exists() or not legacy.exists():
-        return preferred
-    return legacy
+    return WorkspacePaths().state
 
 
 def state_dir_status_payload():
-    preferred = pathlib.Path.cwd() / ".napseer"
-    legacy = pathlib.Path.cwd() / "napseer"
-    active = AUTH_DIR
-    if preferred.exists() and legacy.exists():
-        message = "Both .napseer and legacy napseer directories exist; .napseer is active."
-    elif active == legacy:
-        message = "Legacy napseer directory is active because .napseer does not exist."
-    else:
-        message = ".napseer directory is active."
     return {
-        "active_state_dir": str(active),
-        "preferred_state_dir": str(preferred),
-        "legacy_state_dir": str(legacy),
-        "legacy_state_dir_active": active == legacy,
-        "preferred_state_dir_exists": preferred.exists(),
-        "legacy_state_dir_exists": legacy.exists(),
-        "state_dir_message": message,
+        "active_state_dir": str(AUTH_DIR),
+        "state_dir": str(AUTH_DIR),
+        "state_dir_exists": AUTH_DIR.exists(),
+        "project_root": str(WORKSPACE.root),
+        "state_dir_message": "Private gateway state is stored in user data.",
+        "migration_required": bool(WORKSPACE.migration_sources()),
     }
 
 
@@ -206,7 +194,11 @@ def send_telemetry_event_async(event, component, outcome="success", **fields):
     ).start()
 
 
-AUTH_DIR = local_state_dir()
+WORKSPACE = WorkspacePaths()
+# The pre-dot-directory gateway layout is migration input only.
+if not WORKSPACE.legacy.exists() and any((WORKSPACE.root / "napseer" / name).is_file() for name in ("auth.json", "vault.json")):
+    WORKSPACE.legacy = WORKSPACE.root / "napseer"
+AUTH_DIR = WORKSPACE.state
 AUTH_PATH = AUTH_DIR / "auth.json"
 VAULT_PATH = AUTH_DIR / "vault.json"
 GATEWAY_RELAY_SECRET_PATH = AUTH_DIR / "gateway-relay-secret.json"
@@ -216,10 +208,10 @@ CONTAINER_IDENTITY_PATH = AUTH_DIR / "container-identity.json"
 TELEMETRY_PATH = AUTH_DIR / "telemetry.json"
 INDEX_PATH = AUTH_DIR / "index.sqlite"
 INDEX_LOCK_PATH = AUTH_DIR / "index.lock"
-DEFAULT_KEY_PATH = AUTH_DIR / "id_ed25519"
+DEFAULT_KEY_PATH = WORKSPACE.signing_key
 SCRIPT_PATH = pathlib.Path(__file__).resolve()
 SCRIPT_NAME = SCRIPT_PATH.name
-SCRIPT_DEPENDENCY_NAMES = ("napseer_spake2.py",)
+SCRIPT_DEPENDENCY_NAMES = ("napseer_spake2.py", "napseer_credentials.py")
 CONTRACT_VERSION = "napseer.mcp.contract.v1"
 TELEMETRY_DISABLED_VALUES = {"0", "false", "no", "off", "disabled"}
 TELEMETRY_ALLOWED_FIELDS = {
@@ -687,7 +679,7 @@ def write_vault(passphrase, payload):
     key = _derive_vault_key(passphrase, salt, VAULT_KDF_ITERATIONS)
     plaintext = json.dumps(payload, sort_keys=True, separators=(",", ":")).encode("utf-8")
     nonce, ciphertext, tag = _encrypt_vault_payload(key, plaintext, VAULT_SCHEMA.encode("utf-8"))
-    AUTH_DIR.mkdir(exist_ok=True)
+    AUTH_DIR.mkdir(parents=True, exist_ok=True)
     chmod_best_effort(AUTH_DIR, 0o700)
     VAULT_PATH.write_text(
         json.dumps(
@@ -746,7 +738,7 @@ def derive_gateway_relay_secret(passphrase):
 
 
 def write_gateway_relay_secret(passphrase):
-    AUTH_DIR.mkdir(exist_ok=True)
+    AUTH_DIR.mkdir(parents=True, exist_ok=True)
     chmod_best_effort(AUTH_DIR, 0o700)
     GATEWAY_RELAY_SECRET_PATH.write_text(
         json.dumps(
@@ -802,7 +794,7 @@ def write_gateway_relay_state(**updates):
     current.update({key: value for key, value in updates.items() if value is not None})
     current["schema"] = GATEWAY_RELAY_STATE_SCHEMA
     current["updated_at"] = iso_now()
-    AUTH_DIR.mkdir(exist_ok=True)
+    AUTH_DIR.mkdir(parents=True, exist_ok=True)
     chmod_best_effort(AUTH_DIR, 0o700)
     GATEWAY_RELAY_STATE_PATH.write_text(json.dumps(current, indent=2) + "\n", encoding="utf-8")
     chmod_best_effort(GATEWAY_RELAY_STATE_PATH, 0o600)
@@ -931,7 +923,7 @@ def ensure_container_uuid():
         "container_uuid": value,
         "created_at": iso_now(),
     }
-    AUTH_DIR.mkdir(exist_ok=True)
+    AUTH_DIR.mkdir(parents=True, exist_ok=True)
     chmod_best_effort(AUTH_DIR, 0o700)
     CONTAINER_IDENTITY_PATH.write_text(json.dumps(payload, indent=2, sort_keys=True) + "\n", encoding="utf-8")
     chmod_best_effort(CONTAINER_IDENTITY_PATH, 0o600)
@@ -1164,7 +1156,7 @@ def clean_terminal_name(value, fallback="terminal"):
 
 def default_gateway_session_name():
     source = AUTH.get("project_slug") or AUTH.get("project_name") or pathlib.Path.cwd().name or "project"
-    return f"napseer-{clean_tmux_name(source, 'project')}"
+    return f"napseer-{clean_tmux_name(source, 'project')}-{WORKSPACE.workspace_id[:10]}"
 
 
 def normalize_gateway_command(value=None):
@@ -1727,7 +1719,7 @@ def write_gateway_schedules_state(state):
     state["schema"] = GATEWAY_SCHEDULES_SCHEMA
     state["version"] = 1
     state["updated_at"] = iso_now()
-    AUTH_DIR.mkdir(exist_ok=True)
+    AUTH_DIR.mkdir(parents=True, exist_ok=True)
     chmod_best_effort(AUTH_DIR, 0o700)
     GATEWAY_SCHEDULES_PATH.write_text(json.dumps(state, indent=2, sort_keys=True) + "\n", encoding="utf-8")
     chmod_best_effort(GATEWAY_SCHEDULES_PATH, 0o600)
@@ -5135,7 +5127,7 @@ def request_project_write(method, path, payload, project_id, purpose, scope_type
 
 
 def index_file_lock():
-    AUTH_DIR.mkdir(exist_ok=True)
+    AUTH_DIR.mkdir(parents=True, exist_ok=True)
     chmod_best_effort(AUTH_DIR, 0o700)
     handle = INDEX_LOCK_PATH.open("a+b")
     chmod_best_effort(INDEX_LOCK_PATH, 0o600)
@@ -5153,7 +5145,7 @@ def release_index_file_lock(handle):
 
 
 def index_connect():
-    AUTH_DIR.mkdir(exist_ok=True)
+    AUTH_DIR.mkdir(parents=True, exist_ok=True)
     chmod_best_effort(AUTH_DIR, 0o700)
     conn = sqlite3.connect(str(INDEX_PATH), timeout=30)
     conn.row_factory = sqlite3.Row
@@ -6032,20 +6024,21 @@ def save_auth(updates):
 
 
 def ensure_local_files():
-    AUTH_DIR.mkdir(exist_ok=True)
+    AUTH_DIR.mkdir(parents=True, exist_ok=True)
     chmod_best_effort(AUTH_DIR, 0o700)
     gitignore = AUTH_DIR / ".gitignore"
     if not gitignore.exists():
         gitignore.write_text("*\n!.gitignore\n", encoding="utf-8")
     key_path = pathlib.Path(AUTH["ssh_key_path"] or DEFAULT_KEY_PATH).expanduser()
     public_key_path = pathlib.Path(str(key_path) + ".pub")
-    if not key_path.exists():
-        key_path.parent.mkdir(exist_ok=True)
-        chmod_best_effort(key_path.parent, 0o700)
-        subprocess.run(
-            ["ssh-keygen", "-q", "-t", "ed25519", "-N", "", "-f", str(key_path)],
-            check=True,
-        )
+    with credential_lock(key_path):
+        if not key_path.exists():
+            key_path.parent.mkdir(parents=True, exist_ok=True)
+            chmod_best_effort(key_path.parent, 0o700)
+            subprocess.run(
+                ["ssh-keygen", "-q", "-t", "ed25519", "-N", "", "-f", str(key_path)],
+                check=True,
+            )
     chmod_best_effort(key_path, 0o600)
     if public_key_path.exists():
         chmod_best_effort(public_key_path, 0o644)
@@ -9816,6 +9809,11 @@ def handle(message):
 
 def cli_main(argv):
     command = argv[1] if len(argv) > 1 else ""
+    if argv[1:] == ["auth", "migrate"]:
+        print(json.dumps(WORKSPACE.migrate(gateway=True), indent=2))
+        return
+    if command not in {"help", "-h", "--help"}:
+        WORKSPACE.require_migrated()
     if command in {"configure", "config"}:
         print(json.dumps(cli_configure_status(), indent=2))
         return

scripts/test_gateway_state_paths.py

@@ -0,0 +1,37 @@
+#!/usr/bin/env python3
+"""Exercise gateway imports and its migration command using disposable state."""
+import json
+import os
+from pathlib import Path
+import subprocess
+import sys
+import tempfile
+
+SCRIPT = Path(__file__).resolve().parents[1]/'resources/scripts/napseer_mcp_server.py'
+
+with tempfile.TemporaryDirectory(prefix='gateway-private-state-') as directory:
+    root=Path(directory)/'workspace'; root.mkdir()
+    data=Path(directory)/'data'
+    env=dict(os.environ,NAPSEER_USER_DATA_DIR=str(data),NAPSEER_PROJECT_ROOT=str(root),NAPSEER_TELEMETRY='0')
+    for name in ('NAPSEER_TOKEN','NAPSEER_REFRESH_TOKEN','NAPSEER_API_KEY','NAPSEER_AUTH_FILE'):
+        env.pop(name,None)
+    code="import sys; sys.path.insert(0,sys.argv[1]); import napseer_mcp_server as m; import json; print(json.dumps({'state':str(m.AUTH_DIR),'auth':str(m.AUTH_PATH),'key':str(m.DEFAULT_KEY_PATH)}))"
+    def paths():
+        result=subprocess.run([sys.executable,'-c',code,str(SCRIPT.parent)],cwd=root,env=env,capture_output=True,text=True,timeout=10)
+        assert result.returncode==0,'gateway must load the private-state dependency'
+        return {k:Path(v) for k,v in json.loads(result.stdout).items()}
+    resolved=paths()
+    assert all(path.is_relative_to(data) and not path.is_relative_to(root) for path in resolved.values())
+    legacy=root/'.napseer';legacy.mkdir()
+    (legacy/'auth.json').write_text(json.dumps({'token':'synthetic-worker-token','worker_id':'test-worker'}))
+    (legacy/'id_ed25519').write_bytes(b'synthetic-signing-key')
+    result=subprocess.run([sys.executable,str(SCRIPT),'auth','migrate'],cwd=root,env=env,capture_output=True,text=True,timeout=10)
+    assert result.returncode==0,'gateway migration must complete without enrollment'
+    assert json.loads(result.stdout)['status']=='migrated'
+    assert 'synthetic-worker-token' not in result.stdout+result.stderr
+    assert json.loads(resolved['auth'].read_text())['worker_id']=='test-worker'
+    assert resolved['key'].read_bytes()==b'synthetic-signing-key'
+    assert not list(legacy.iterdir())
+    assert paths()==resolved
+
+print('ok: gateway private state and migration stay outside the checkout')