diff --git a/leapp/snactor/commands/discover.py b/leapp/snactor/commands/discover.py
index 5c7180bb4..00429582e 100644
--- a/leapp/snactor/commands/discover.py
+++ b/leapp/snactor/commands/discover.py
@@ -10,6 +10,7 @@
 from leapp.utils.repository import requires_repository, find_repository_basedir, get_repository_name
 from leapp.utils.clicmd import command, command_opt
 from leapp.workflows import get_workflows
+from leapp.snactor.utils import safe_discover
 
 
 def _is_local(repository, cls, base_dir, all_repos=False):
@@ -85,9 +86,22 @@ def _get_model_details(model):
          description=_LONG_DESCRIPTION)
 @command_opt('json', is_flag=True, help='Output in json format instead of human readable form')
 @command_opt('all', is_flag=True, help='Include items from linked repositories')
+@command_opt('safe', is_flag=True, help='Analyze actor files statically to work around runtime errors')
 @requires_repository
 def cli(args):
     base_dir = find_repository_basedir('.')
+
+    if args.safe and args.json:
+        sys.stderr.write('The options --safe and --json are currently mutually exclusive\n')
+        sys.exit(1)
+
+    if args.safe:
+        sys.stdout.write(
+            'Repository:\n  Name: {repository}\n  Path: {base_dir}\n\n'.format(repository=get_repository_name(base_dir),
+                                                                               base_dir=base_dir))
+        safe_discover(base_dir)
+        sys.exit(0)
+
     repository = find_and_scan_repositories(base_dir, include_locals=True)
     try:
         repository.load()
diff --git a/leapp/snactor/utils.py b/leapp/snactor/utils.py
new file mode 100644
index 000000000..83887efb1
--- /dev/null
+++ b/leapp/snactor/utils.py
@@ -0,0 +1,128 @@
+import ast
+import functools
+import itertools
+import os
+import six
+
+
+def print_section(data, section, pivot):
+    "Print a section of obtained data"
+    type_data = data[section]
+    print('{}'.format(section.capitalize()))
+    for td in type_data:
+        fp = format_file_path(pivot, td['file'])
+        first_part = '  - {}({})'.format(td['name'], ', '.join(td['bases']))
+        pad = '.' * (60 - len(first_part))
+        print('{} {} {}'.format(first_part, pad, fp))
+    print('')
+
+
+def format_file_path(pivot, path):
+    "Format path as relative to a pivot"
+    if not pivot or pivot == '.':
+        pivot = os.getcwd()
+    return os.path.relpath(path, pivot)
+
+
+def get_candidate_files(start='.'):
+    "Find all .py files in a directory tree"
+    for root, _, files in os.walk(start):
+        for f in files:
+            if not f.endswith('py'):
+                continue
+            yield os.path.join(root, f)
+
+
+def ast_parse_file(file):
+    "Parse a python file and return tuple (ast, filename)"
+    with open(file, mode='r') as fp:
+        try:
+            return ast.parse(fp.read(), file), file
+        except Exception:
+            return None, file
+
+
+def get_base_classes(bases, via):
+    "Get base classes of a type, only direct names are supported currently"
+    bases_set = set()
+    errors = []
+    for base in bases:
+        if isinstance(base, ast.Name):
+            bases_set.add(base.id)
+        else:
+            errors.append('Unknown base: {} via {}'.format(base.__class__.__name__, via))
+    return bases_set, errors
+
+
+def inspect(tree_file, collected_types={}, type_infos={}):
+    "Inspect and collect data from AST tree"
+    tree, file = tree_file
+    if not tree:
+        return ['Unable to parse: {}'.format(file)]
+    errors = []
+    for node in ast.walk(tree):
+        if isinstance(node, ast.ClassDef):
+            base_classes, err = get_base_classes(node.bases, file)
+            errors += err
+
+            if base_classes & collected_types['models']:
+                collected_types['models'].add(node.name)
+                type_infos['models'].append({
+                    'name': node.name,
+                    'bases': list(base_classes),
+                    'file': file
+                })
+            if base_classes & collected_types['actors']:
+                collected_types['actors'].add(node.name)
+                type_infos['actors'].append({
+                    'name': node.name,
+                    'bases': list(base_classes),
+                    'file': file
+                })
+            if base_classes & collected_types['tags']:
+                collected_types['tags'].add(node.name)
+                type_infos['tags'].append({
+                    'name': node.name,
+                    'bases': list(base_classes),
+                    'file': file
+                })
+    return errors
+
+
+def safe_discover(pivot):
+    # Here we collect all the types that inherit from Model/Actor/Tag types
+    # so that we can use this dict as a lookup for deeper search to support
+    # use cases like:
+    #
+    # class FirstOrderModel(Model):
+    #     pass
+    #
+    # class SecondOrderModel(FirstOrderModel):
+    #     pass
+    #
+    collected_types = {
+        'models': set(['Model']),
+        'actors': set(['Actor']),
+        'tags': set(['Tag'])
+    }
+    type_infos = {
+        'models': [],
+        'actors': [],
+        'tags': []
+    }
+
+    inspector = functools.partial(
+        inspect,
+        collected_types=collected_types,
+        type_infos=type_infos
+    )
+
+    errors = filter(None, map(inspector, map(ast_parse_file, get_candidate_files(pivot))))
+    flat_errors = list(itertools.chain(*errors))
+
+    print_section(type_infos, 'actors', pivot)
+    print_section(type_infos, 'models', pivot)
+    print_section(type_infos, 'tags', pivot)
+    if flat_errors:
+        print('Errors:')
+        print('\n'.join(flat_errors))