feat(update): auto-backup HERMES_HOME before hermes update (NousResearch#16539)

Every 'hermes update' now runs a full backup of ~/.hermes/ first, so
users can always roll back to the exact state they had before the
update if anything goes wrong (corrupted sessions.db, broken skills,
config migrations that don't round-trip, etc.).

Changes:
- hermes_cli/backup.py: new create_pre_update_backup() helper. Writes
  to <HERMES_HOME>/backups/pre-update-<stamp>.zip using the same
  exclusion rules and SQLite safe-copy as 'hermes backup'. Auto-rotates
  (keep last N, pre-update-*.zip only — hand-dropped zips in backups/
  are untouched). Adds 'backups' to _EXCLUDED_DIRS so subsequent backups
  don't nest prior ones.
- hermes_cli/main.py: _run_pre_update_backup() wired into
  _cmd_update_impl before any git operation. Prints save path, restore
  command, and how to disable. Swallows failures so a broken backup
  never blocks the update itself. New --no-backup flag on 'hermes
  update' for one-off override.
- hermes_cli/config.py: new 'updates' section in DEFAULT_CONFIG with
  pre_update_backup (default true) and backup_keep (default 5).
  Auto-surfaces in the dashboard config UI.
- tests/hermes_cli/test_backup.py: +11 tests covering backup location,
  content parity with 'hermes backup', no-recursion, rotation, manual
  file preservation, config gate, --no-backup flag, flag-wins-over-config.

Author: teknium1

hermes_cli/backup.py

@@ -36,6 +36,7 @@
     "__pycache__",      # bytecode caches — regenerated on import
     ".git",             # nested git dirs (profiles shouldn't have these, but safety)
     "node_modules",     # js deps if website/ somehow leaks in
+    "backups",          # prior auto-backups — don't nest backups exponentially
 }
 
 # File-name suffixes to skip
@@ -683,3 +684,138 @@ def run_quick_backup(args) -> None:
         print(f"  Restore with: /snapshot restore {snap_id}")
     else:
         print("No state files found to snapshot.")
+
+
+# ---------------------------------------------------------------------------
+# Pre-update auto-backup
+# ---------------------------------------------------------------------------
+
+_PRE_UPDATE_BACKUPS_DIR = "backups"
+_PRE_UPDATE_PREFIX = "pre-update-"
+_PRE_UPDATE_DEFAULT_KEEP = 5
+
+
+def _pre_update_backup_dir(hermes_home: Optional[Path] = None) -> Path:
+    home = hermes_home or get_hermes_home()
+    return home / _PRE_UPDATE_BACKUPS_DIR
+
+
+def _prune_pre_update_backups(backup_dir: Path, keep: int) -> int:
+    """Remove oldest pre-update backups beyond the keep limit.
+
+    Returns the number of files deleted.  Only touches files matching
+    ``pre-update-*.zip`` so hand-made zips dropped in the same directory
+    are never touched.
+    """
+    if keep < 0:
+        keep = 0
+    if not backup_dir.exists():
+        return 0
+
+    backups = sorted(
+        (p for p in backup_dir.iterdir()
+         if p.is_file() and p.name.startswith(_PRE_UPDATE_PREFIX) and p.suffix.lower() == ".zip"),
+        key=lambda p: p.name,
+        reverse=True,
+    )
+
+    deleted = 0
+    for p in backups[keep:]:
+        try:
+            p.unlink()
+            deleted += 1
+        except OSError as exc:
+            logger.warning("Failed to prune backup %s: %s", p.name, exc)
+
+    return deleted
+
+
+def create_pre_update_backup(
+    hermes_home: Optional[Path] = None,
+    keep: int = _PRE_UPDATE_DEFAULT_KEEP,
+) -> Optional[Path]:
+    """Create a full zip backup of HERMES_HOME under ``backups/``.
+
+    Mirrors :func:`run_backup` (same exclusion rules, same SQLite safe-copy)
+    but writes to ``<HERMES_HOME>/backups/pre-update-<timestamp>.zip`` and
+    auto-prunes old pre-update backups.
+
+    Returns the path to the created zip, or ``None`` if no files were
+    found or the backup could not be created.  Never raises — the caller
+    (``hermes update``) should continue even if the backup fails.
+    """
+    hermes_root = hermes_home or get_default_hermes_root()
+    if not hermes_root.is_dir():
+        return None
+
+    backup_dir = _pre_update_backup_dir(hermes_root)
+    try:
+        backup_dir.mkdir(parents=True, exist_ok=True)
+    except OSError as exc:
+        logger.warning("Could not create pre-update backup dir %s: %s", backup_dir, exc)
+        return None
+
+    stamp = datetime.now().strftime("%Y-%m-%d-%H%M%S")
+    out_path = backup_dir / f"{_PRE_UPDATE_PREFIX}{stamp}.zip"
+
+    # Collect files (same logic as run_backup, minus the chatty progress prints)
+    files_to_add: list[tuple[Path, Path]] = []
+    try:
+        for dirpath, dirnames, filenames in os.walk(hermes_root, followlinks=False):
+            dp = Path(dirpath)
+            # Prune excluded directories in-place so os.walk doesn't descend
+            dirnames[:] = [d for d in dirnames if d not in _EXCLUDED_DIRS]
+
+            for fname in filenames:
+                fpath = dp / fname
+                try:
+                    rel = fpath.relative_to(hermes_root)
+                except ValueError:
+                    continue
+
+                if _should_exclude(rel):
+                    continue
+
+                # Skip the output zip itself if it already exists
+                try:
+                    if fpath.resolve() == out_path.resolve():
+                        continue
+                except (OSError, ValueError):
+                    pass
+
+                files_to_add.append((fpath, rel))
+    except OSError as exc:
+        logger.warning("Pre-update backup: walk failed: %s", exc)
+        return None
+
+    if not files_to_add:
+        return None
+
+    try:
+        with zipfile.ZipFile(out_path, "w", zipfile.ZIP_DEFLATED, compresslevel=6) as zf:
+            for abs_path, rel_path in files_to_add:
+                try:
+                    if abs_path.suffix == ".db":
+                        with tempfile.NamedTemporaryFile(suffix=".db", delete=False) as tmp:
+                            tmp_db = Path(tmp.name)
+                        try:
+                            if _safe_copy_db(abs_path, tmp_db):
+                                zf.write(tmp_db, arcname=str(rel_path))
+                        finally:
+                            tmp_db.unlink(missing_ok=True)
+                    else:
+                        zf.write(abs_path, arcname=str(rel_path))
+                except (PermissionError, OSError, ValueError) as exc:
+                    logger.debug("Skipping %s in pre-update backup: %s", rel_path, exc)
+                    continue
+    except OSError as exc:
+        logger.warning("Pre-update backup: zip write failed: %s", exc)
+        # Best-effort cleanup of partial file
+        try:
+            out_path.unlink(missing_ok=True)
+        except OSError:
+            pass
+        return None
+
+    _prune_pre_update_backups(backup_dir, keep=keep)
+    return out_path

hermes_cli/config.py

@@ -1037,6 +1037,19 @@ def _ensure_hermes_home_managed(home: Path):
         "seen": {},
     },
 
+    # ``hermes update`` behaviour.
+    "updates": {
+        # Run a full ``hermes backup``-style zip of HERMES_HOME before every
+        # ``hermes update``.  Backups land in ``<HERMES_HOME>/backups/`` and
+        # can be restored with ``hermes import <path>``.  Set to false to
+        # skip the backup entirely; use the ``--no-backup`` flag on a single
+        # update invocation to override just that run.
+        "pre_update_backup": True,
+        # How many pre-update backup zips to retain.  Older ones are pruned
+        # automatically after each successful backup.
+        "backup_keep": 5,
+    },
+
     # Config schema version - bump this when adding new required fields
     "_config_version": 22,
 }

hermes_cli/main.py

@@ -6142,6 +6142,90 @@ def _ensure_fhs_path_guard() -> None:
         print("    (reload your shell or run 'source ~/.bashrc' to pick it up)")
 
 
+def _run_pre_update_backup(args) -> None:
+    """Create a full zip backup of HERMES_HOME before running the update.
+
+    Gated on ``updates.pre_update_backup`` in config (default true).  The
+    ``--no-backup`` flag on ``hermes update`` overrides it for one run.
+    Never raises — a backup failure should not block the update itself.
+    """
+    # CLI flag wins over config
+    if getattr(args, "no_backup", False):
+        print("◆ Pre-update backup: skipped (--no-backup)")
+        print()
+        return
+
+    try:
+        from hermes_cli.config import load_config
+        cfg = load_config()
+    except Exception as exc:
+        logging.getLogger(__name__).debug("Could not load config for pre-update backup: %s", exc)
+        cfg = {}
+
+    updates_cfg = cfg.get("updates", {}) if isinstance(cfg, dict) else {}
+    enabled = updates_cfg.get("pre_update_backup", True)
+    keep = updates_cfg.get("backup_keep", 5)
+
+    if not enabled:
+        print("◆ Pre-update backup: disabled (updates.pre_update_backup=false in config.yaml)")
+        print()
+        return
+
+    try:
+        from hermes_cli.backup import create_pre_update_backup
+    except Exception as exc:
+        print(f"⚠ Pre-update backup: could not load backup module ({exc}); continuing update.")
+        print()
+        return
+
+    print("◆ Creating pre-update backup...")
+    t0 = _time.monotonic()
+    try:
+        out_path = create_pre_update_backup(keep=int(keep))
+    except Exception as exc:  # defensive — helper already swallows, but just in case
+        print(f"  ⚠ Backup failed: {exc}")
+        print("  Continuing with update.")
+        print()
+        return
+
+    elapsed = _time.monotonic() - t0
+
+    if out_path is None:
+        print("  ⚠ Backup skipped (no files found or write failed); continuing update.")
+        print()
+        return
+
+    try:
+        size_bytes = out_path.stat().st_size
+    except OSError:
+        size_bytes = 0
+
+    # Human-readable size
+    size_str = f"{size_bytes} B"
+    for unit in ("KB", "MB", "GB"):
+        if size_bytes < 1024:
+            break
+        size_bytes /= 1024
+        size_str = f"{size_bytes:.1f} {unit}"
+
+    # Render path using display_hermes_home so the user sees ~/.hermes/...
+    try:
+        from hermes_constants import get_hermes_home, display_hermes_home
+        home = get_hermes_home()
+        try:
+            display_path = f"{display_hermes_home()}/{out_path.relative_to(home)}"
+        except ValueError:
+            display_path = str(out_path)
+    except Exception:
+        display_path = str(out_path)
+
+    print(f"  Saved:    {display_path} ({size_str}, {elapsed:.1f}s)")
+    print(f"  Restore:  hermes import {out_path}")
+    print(f"  Disable:  set updates.pre_update_backup: false in config.yaml")
+    print(f"            (or pass --no-backup on a single update)")
+    print()
+
+
 def cmd_update(args):
     """Update Hermes Agent to the latest version.
 
@@ -6184,6 +6268,10 @@ def _cmd_update_impl(args, gateway_mode: bool):
     print("⚕ Updating Hermes Agent...")
     print()
 
+    # Pre-update backup — runs before any git/file mutation so users can
+    # always roll back to the exact state they had before this update.
+    _run_pre_update_backup(args)
+
     # Try git-based update first, fall back to ZIP download on Windows
     # when git file I/O is broken (antivirus, NTFS filter drivers, etc.)
     use_zip_update = False
@@ -9577,6 +9665,12 @@ def cmd_claw(args):
         default=False,
         help="Check whether an update is available without installing anything",
     )
+    update_parser.add_argument(
+        "--no-backup",
+        action="store_true",
+        default=False,
+        help="Skip the pre-update backup for this run (overrides updates.pre_update_backup)",
+    )
     update_parser.set_defaults(func=cmd_update)
 
     # =========================================================================