diff --git a/doc/source/engines.rst b/doc/source/engines.rst
index 6aa2388f4..312c2e367 100644
--- a/doc/source/engines.rst
+++ b/doc/source/engines.rst
@@ -448,11 +448,13 @@ Components
 .. automodule:: taskflow.engines.action_engine.builder
 .. automodule:: taskflow.engines.action_engine.compiler
 .. automodule:: taskflow.engines.action_engine.completer
+.. automodule:: taskflow.engines.action_engine.deciders
 .. automodule:: taskflow.engines.action_engine.executor
 .. automodule:: taskflow.engines.action_engine.runtime
 .. automodule:: taskflow.engines.action_engine.scheduler
 .. autoclass:: taskflow.engines.action_engine.scopes.ScopeWalker
     :special-members: __iter__
+.. automodule:: taskflow.engines.action_engine.traversal
 
 Hierarchy
 =========
diff --git a/doc/source/patterns.rst b/doc/source/patterns.rst
index 8c8eb410e..fc546bb37 100644
--- a/doc/source/patterns.rst
+++ b/doc/source/patterns.rst
@@ -21,6 +21,7 @@ Graph flow
 ~~~~~~~~~~
 
 .. automodule:: taskflow.patterns.graph_flow
+.. automodule:: taskflow.deciders
 
 Hierarchy
 ~~~~~~~~~
diff --git a/taskflow/deciders.py b/taskflow/deciders.py
new file mode 100644
index 000000000..c96b32185
--- /dev/null
+++ b/taskflow/deciders.py
@@ -0,0 +1,99 @@
+# -*- coding: utf-8 -*-
+
+#    Copyright (C) 2012 Yahoo! Inc. All Rights Reserved.
+#
+#    Licensed under the Apache License, Version 2.0 (the "License"); you may
+#    not use this file except in compliance with the License. You may obtain
+#    a copy of the License at
+#
+#         http://www.apache.org/licenses/LICENSE-2.0
+#
+#    Unless required by applicable law or agreed to in writing, software
+#    distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
+#    WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
+#    License for the specific language governing permissions and limitations
+#    under the License.
+
+import six
+
+from taskflow.utils import misc
+
+
+class Depth(misc.StrEnum):
+    """Enumeration of decider(s) *area of influence*."""
+
+    ALL = 'ALL'
+    """
+    **Default** decider depth that affects **all** successor atoms (including
+    ones that are in successor nested flows).
+    """
+
+    FLOW = 'FLOW'
+    """
+    Decider depth that affects **all** successor tasks in the **same**
+    flow (it will **not** affect tasks/retries that are in successor
+    nested flows).
+
+    .. warning::
+
+       While using this kind we are allowed to execute successors of
+       things that have been ignored (for example nested flows and the
+       tasks they contain), this may result in symbol lookup errors during
+       running, user beware.
+    """
+
+    NEIGHBORS = 'NEIGHBORS'
+    """
+    Decider depth that affects only **next** successor tasks (and does
+    not traverse past **one** level of successor tasks).
+
+    .. warning::
+
+       While using this kind we are allowed to execute successors of
+       things that have been ignored (for example nested flows and the
+       tasks they contain), this may result in symbol lookup errors during
+       running, user beware.
+    """
+
+    ATOM = 'ATOM'
+    """
+    Decider depth that affects only **targeted** atom (and does
+    **not** traverse into **any** level of successor atoms).
+
+    .. warning::
+
+       While using this kind we are allowed to execute successors of
+       things that have been ignored (for example nested flows and the
+       tasks they contain), this may result in symbol lookup errors during
+       running, user beware.
+    """
+
+    @classmethod
+    def translate(cls, desired_depth):
+        """Translates a string into a depth enumeration."""
+        if isinstance(desired_depth, cls):
+            # Nothing to do in the first place...
+            return desired_depth
+        if not isinstance(desired_depth, six.string_types):
+            raise TypeError("Unexpected desired depth type, string type"
+                            " expected, not %s" % type(desired_depth))
+        try:
+            return cls(desired_depth.upper())
+        except ValueError:
+            pretty_depths = sorted([a_depth.name for a_depth in cls])
+            raise ValueError("Unexpected decider depth value, one of"
+                             " %s (case-insensitive) is expected and"
+                             " not '%s'" % (pretty_depths, desired_depth))
+
+
+# Depth area of influence order (from greater influence to least).
+#
+# Order very much matters here...
+_ORDERING = tuple([
+    Depth.ALL, Depth.FLOW, Depth.NEIGHBORS, Depth.ATOM,
+])
+
+
+def pick_widest(depths):
+    """Pick from many depths which has the **widest** area of influence."""
+    return _ORDERING[min(_ORDERING.index(d) for d in depths)]
diff --git a/taskflow/engines/action_engine/analyzer.py b/taskflow/engines/action_engine/analyzer.py
index c0b625b64..e7047c607 100644
--- a/taskflow/engines/action_engine/analyzer.py
+++ b/taskflow/engines/action_engine/analyzer.py
@@ -14,108 +14,16 @@
 #    License for the specific language governing permissions and limitations
 #    under the License.
 
-import abc
-import itertools
 import operator
 import weakref
 
-import six
-
 from taskflow.engines.action_engine import compiler as co
+from taskflow.engines.action_engine import deciders
+from taskflow.engines.action_engine import traversal
 from taskflow import states as st
 from taskflow.utils import iter_utils
 
 
-def _depth_first_iterate(graph, connected_to_functors, initial_nodes_iter):
-    """Iterates connected nodes in execution graph (from starting set).
-
-    Jumps over nodes with ``noop`` attribute (does not yield them back).
-    """
-    stack = list(initial_nodes_iter)
-    while stack:
-        node = stack.pop()
-        node_attrs = graph.node[node]
-        if not node_attrs.get('noop'):
-            yield node
-        try:
-            node_kind = node_attrs['kind']
-            connected_to_functor = connected_to_functors[node_kind]
-        except KeyError:
-            pass
-        else:
-            stack.extend(connected_to_functor(node))
-
-
-@six.add_metaclass(abc.ABCMeta)
-class Decider(object):
-    """Base class for deciders.
-
-    Provides interface to be implemented by sub-classes
-    Decider checks whether next atom in flow should be executed or not
-    """
-
-    @abc.abstractmethod
-    def check(self, runtime):
-        """Returns bool of whether this decider should allow running."""
-
-    @abc.abstractmethod
-    def affect(self, runtime):
-        """If the :py:func:`~.check` returns false, affects associated atoms.
-
-        """
-
-    def check_and_affect(self, runtime):
-        """Handles :py:func:`~.check` + :py:func:`~.affect` in right order."""
-        proceed = self.check(runtime)
-        if not proceed:
-            self.affect(runtime)
-        return proceed
-
-
-class IgnoreDecider(Decider):
-    """Checks any provided edge-deciders and determines if ok to run."""
-
-    def __init__(self, atom, edge_deciders):
-        self._atom = atom
-        self._edge_deciders = edge_deciders
-
-    def check(self, runtime):
-        """Returns bool of whether this decider should allow running."""
-        # Gather all atoms results so that those results can be used
-        # by the decider(s) that are making a decision as to pass or
-        # not pass...
-        results = {}
-        for node, node_kind, _local_decider in self._edge_deciders:
-            if node_kind in co.ATOMS:
-                results[node.name] = runtime.storage.get(node.name)
-        for _node, _node_kind, local_decider in self._edge_deciders:
-            if not local_decider(history=results):
-                return False
-        return True
-
-    def affect(self, runtime):
-        """If the :py:func:`~.check` returns false, affects associated atoms.
-
-        This will alter the associated atom + successor atoms by setting there
-        state to ``IGNORE`` so that they are ignored in future runtime
-        activities.
-        """
-        successors_iter = runtime.analyzer.iterate_connected_atoms(self._atom)
-        runtime.reset_atoms(itertools.chain([self._atom], successors_iter),
-                            state=st.IGNORE, intention=st.IGNORE)
-
-
-class NoOpDecider(Decider):
-    """No-op decider that says it is always ok to run & has no effect(s)."""
-
-    def check(self, runtime):
-        """Always good to go."""
-        return True
-
-    def affect(self, runtime):
-        """Does nothing."""
-
-
 class Analyzer(object):
     """Analyzes a compilation and aids in execution processes.
 
@@ -142,7 +50,7 @@ class Analyzer(object):
         if state == st.SUCCESS:
             if intention == st.REVERT:
                 return iter([
-                    (atom, NoOpDecider()),
+                    (atom, deciders.NoOpDecider()),
                 ])
             elif intention == st.EXECUTE:
                 return self.browse_atoms_for_execute(atom=atom)
@@ -165,10 +73,15 @@ class Analyzer(object):
         if atom is None:
             atom_it = self.iterate_nodes(co.ATOMS)
         else:
-            successors_iter = self._execution_graph.successors_iter
-            atom_it = _depth_first_iterate(self._execution_graph,
-                                           {co.FLOW: successors_iter},
-                                           successors_iter(atom))
+            # NOTE(harlowja): the reason this uses breadth first is so that
+            # when deciders are applied that those deciders can be applied
+            # from top levels to lower levels since lower levels *may* be
+            # able to run even if top levels have deciders that decide to
+            # ignore some atoms... (going deeper first would make this
+            # problematic to determine as top levels can have their deciders
+            # applied **after** going deeper).
+            atom_it = traversal.breadth_first_iterate(
+                self._execution_graph, atom, traversal.Direction.FORWARD)
         for atom in atom_it:
             is_ready, late_decider = self._get_maybe_ready_for_execute(atom)
             if is_ready:
@@ -185,18 +98,50 @@ class Analyzer(object):
         if atom is None:
             atom_it = self.iterate_nodes(co.ATOMS)
         else:
-            predecessors_iter = self._execution_graph.predecessors_iter
-            atom_it = _depth_first_iterate(self._execution_graph,
-                                           {co.FLOW: predecessors_iter},
-                                           predecessors_iter(atom))
+            atom_it = traversal.breadth_first_iterate(
+                self._execution_graph, atom, traversal.Direction.BACKWARD,
+                # Stop at the retry boundary (as retries 'control' there
+                # surronding atoms, and we don't want to back track over
+                # them so that they can correctly affect there associated
+                # atoms); we do though need to jump through all tasks since
+                # if a predecessor Y was ignored and a predecessor Z before Y
+                # was not it should be eligible to now revert...
+                through_retries=False)
         for atom in atom_it:
             is_ready, late_decider = self._get_maybe_ready_for_revert(atom)
             if is_ready:
                 yield (atom, late_decider)
 
     def _get_maybe_ready(self, atom, transition_to, allowed_intentions,
-                         connected_fetcher, connected_checker,
+                         connected_fetcher, ready_checker,
                          decider_fetcher):
+        def iter_connected_states():
+            # Lazily iterate over connected states so that ready checkers
+            # can stop early (vs having to consume and check all the
+            # things...)
+            for atom in connected_fetcher():
+                # TODO(harlowja): make this storage api better, its not
+                # especially clear what the following is doing (mainly
+                # to avoid two calls into storage).
+                atom_states = self._storage.get_atoms_states([atom.name])
+                yield (atom, atom_states[atom.name])
+        # NOTE(harlowja): How this works is the following...
+        #
+        # 1. First check if the current atom can even transition to the
+        #    desired state, if not this atom is definitely not ready to
+        #    execute or revert.
+        # 2. Check if the actual atoms intention is in one of the desired/ok
+        #    intentions, if it is not there we are still not ready to execute
+        #    or revert.
+        # 3. Iterate over (atom, atom_state, atom_intention) for all the
+        #    atoms the 'connected_fetcher' callback yields from underlying
+        #    storage and direct that iterator into the 'ready_checker'
+        #    callback, that callback should then iterate over these entries
+        #    and determine if it is ok to execute or revert.
+        # 4. If (and only if) 'ready_checker' returns true, then
+        #    the 'decider_fetcher' callback is called to get a late decider
+        #    which can (if it desires) affect this ready result (but does
+        #    so right before the atom is about to be scheduled).
         state = self._storage.get_atom_state(atom.name)
         ok_to_transition = self._runtime.check_atom_transition(atom, state,
                                                                transition_to)
@@ -205,59 +150,62 @@ class Analyzer(object):
         intention = self._storage.get_atom_intention(atom.name)
         if intention not in allowed_intentions:
             return (False, None)
-        connected_states = self._storage.get_atoms_states(
-            connected_atom.name for connected_atom in connected_fetcher(atom))
-        ok_to_run = connected_checker(six.itervalues(connected_states))
+        ok_to_run = ready_checker(iter_connected_states())
         if not ok_to_run:
             return (False, None)
         else:
-            return (True, decider_fetcher(atom))
+            return (True, decider_fetcher())
 
     def _get_maybe_ready_for_execute(self, atom):
         """Returns if an atom is *likely* ready to be executed."""
-        def decider_fetcher(atom):
-            edge_deciders = self._runtime.fetch_edge_deciders(atom)
-            if edge_deciders:
-                return IgnoreDecider(atom, edge_deciders)
-            else:
-                return NoOpDecider()
-        predecessors_iter = self._execution_graph.predecessors_iter
-        connected_fetcher = lambda atom: \
-            _depth_first_iterate(self._execution_graph,
-                                 {co.FLOW: predecessors_iter},
-                                 predecessors_iter(atom))
-        connected_checker = lambda connected_iter: \
-            all(state == st.SUCCESS and intention == st.EXECUTE
-                for state, intention in connected_iter)
+        def ready_checker(pred_connected_it):
+            for _atom, (atom_state, atom_intention) in pred_connected_it:
+                if (atom_state in (st.SUCCESS, st.IGNORE) and
+                        atom_intention in (st.EXECUTE, st.IGNORE)):
+                    continue
+                return False
+            return True
+        decider_fetcher = lambda: \
+            deciders.IgnoreDecider(
+                atom, self._runtime.fetch_edge_deciders(atom))
+        connected_fetcher = lambda: \
+            traversal.depth_first_iterate(self._execution_graph, atom,
+                                          # Whether the desired atom
+                                          # can execute is dependent on its
+                                          # predecessors outcomes (thus why
+                                          # we look backwards).
+                                          traversal.Direction.BACKWARD)
+        # If this atoms current state is able to be transitioned to RUNNING
+        # and its intention is to EXECUTE and all of its predecessors executed
+        # successfully or were ignored then this atom is ready to execute.
         return self._get_maybe_ready(atom, st.RUNNING, [st.EXECUTE],
-                                     connected_fetcher, connected_checker,
+                                     connected_fetcher, ready_checker,
                                      decider_fetcher)
 
     def _get_maybe_ready_for_revert(self, atom):
         """Returns if an atom is *likely* ready to be reverted."""
-        successors_iter = self._execution_graph.successors_iter
-        connected_fetcher = lambda atom: \
-            _depth_first_iterate(self._execution_graph,
-                                 {co.FLOW: successors_iter},
-                                 successors_iter(atom))
-        connected_checker = lambda connected_iter: \
-            all(state in (st.PENDING, st.REVERTED)
-                for state, _intention in connected_iter)
-        decider_fetcher = lambda atom: NoOpDecider()
+        def ready_checker(succ_connected_it):
+            for _atom, (atom_state, _atom_intention) in succ_connected_it:
+                if atom_state not in (st.PENDING, st.REVERTED, st.IGNORE):
+                    return False
+            return True
+        noop_decider = deciders.NoOpDecider()
+        connected_fetcher = lambda: \
+            traversal.depth_first_iterate(self._execution_graph, atom,
+                                          # Whether the desired atom
+                                          # can revert is dependent on its
+                                          # successors states (thus why we
+                                          # look forwards).
+                                          traversal.Direction.FORWARD)
+        decider_fetcher = lambda: noop_decider
+        # If this atoms current state is able to be transitioned to REVERTING
+        # and its intention is either REVERT or RETRY and all of its
+        # successors are either PENDING or REVERTED then this atom is ready
+        # to revert.
         return self._get_maybe_ready(atom, st.REVERTING, [st.REVERT, st.RETRY],
-                                     connected_fetcher, connected_checker,
+                                     connected_fetcher, ready_checker,
                                      decider_fetcher)
 
-    def iterate_connected_atoms(self, atom):
-        """Iterates **all** successor atoms connected to given atom."""
-        successors_iter = self._execution_graph.successors_iter
-        return _depth_first_iterate(
-            self._execution_graph, {
-                co.FLOW: successors_iter,
-                co.TASK: successors_iter,
-                co.RETRY: successors_iter,
-            }, successors_iter(atom))
-
     def iterate_retries(self, state=None):
         """Iterates retry atoms that match the provided state.
 
@@ -268,7 +216,8 @@ class Analyzer(object):
             atom_states = self._storage.get_atoms_states(atom.name
                                                          for atom in atoms)
             for atom in atoms:
-                if atom_states[atom.name][0] == state:
+                atom_state, _atom_intention = atom_states[atom.name]
+                if atom_state == state:
                     yield atom
         else:
             for atom in self.iterate_nodes((co.RETRY,)):
@@ -290,7 +239,7 @@ class Analyzer(object):
         atom_states = self._storage.get_atoms_states(atom.name
                                                      for atom in atoms)
         for atom in atoms:
-            atom_state = atom_states[atom.name][0]
+            atom_state, _atom_intention = atom_states[atom.name]
             if atom_state == st.IGNORE:
                 continue
             if atom_state != st.SUCCESS:
diff --git a/taskflow/engines/action_engine/builder.py b/taskflow/engines/action_engine/builder.py
index 72aa81c56..7ce9b301a 100644
--- a/taskflow/engines/action_engine/builder.py
+++ b/taskflow/engines/action_engine/builder.py
@@ -223,11 +223,15 @@ class MachineBuilder(object):
                                           atom, intention)
                 except Exception:
                     memory.failures.append(failure.Failure())
+                    LOG.exception("Engine '%s' atom post-completion"
+                                  " failed", atom)
                 else:
                     try:
                         more_work = set(iter_next_atoms(atom=atom))
                     except Exception:
                         memory.failures.append(failure.Failure())
+                        LOG.exception("Engine '%s' atom post-completion"
+                                      " next atom searching failed", atom)
                     else:
                         next_up.update(more_work)
             if is_runnable() and next_up and not memory.failures:
diff --git a/taskflow/engines/action_engine/compiler.py b/taskflow/engines/action_engine/compiler.py
index 561fc18ce..9933fd0cb 100644
--- a/taskflow/engines/action_engine/compiler.py
+++ b/taskflow/engines/action_engine/compiler.py
@@ -40,22 +40,55 @@ LOG = logging.getLogger(__name__)
 TASK = 'task'
 RETRY = 'retry'
 FLOW = 'flow'
+FLOW_END = 'flow_end'
 
 # Quite often used together, so make a tuple everyone can share...
 ATOMS = (TASK, RETRY)
+FLOWS = (FLOW, FLOW_END)
+
+
+class Terminator(object):
+    """Flow terminator class."""
+
+    def __init__(self, flow):
+        self._flow = flow
+        self._name = "%s[$]" % (self._flow.name,)
+
+    @property
+    def flow(self):
+        """The flow which this terminator signifies/marks the end of."""
+        return self._flow
+
+    @property
+    def name(self):
+        """Useful name this end terminator has (derived from flow name)."""
+        return self._name
+
+    def __str__(self):
+        return "%s[$]" % (self._flow,)
 
 
 class Compilation(object):
-    """The result of a compilers compile() is this *immutable* object."""
+    """The result of a compilers ``compile()`` is this *immutable* object."""
 
-    #: Task nodes will have a ``kind`` attribute/metadata key with this value.
+    #: Task nodes will have a ``kind`` metadata key with this value.
     TASK = TASK
 
-    #: Retry nodes will have a ``kind`` attribute/metadata key with this value.
+    #: Retry nodes will have a ``kind`` metadata key with this value.
     RETRY = RETRY
 
-    #: Flow nodes will have a ``kind`` attribute/metadata key with this value.
     FLOW = FLOW
+    """
+    Flow **entry** nodes will have a ``kind`` metadata key with
+    this value.
+    """
+
+    FLOW_END = FLOW_END
+    """
+    Flow **exit** nodes will have a ``kind`` metadata key with
+    this value (only applicable for compilation execution graph, not currently
+    used in tree hierarchy).
+    """
 
     def __init__(self, execution_graph, hierarchy):
         self._execution_graph = execution_graph
@@ -141,6 +174,8 @@ class FlowCompiler(object):
             _add_update_edges(graph, u_graph.no_successors_iter(),
                               list(v_graph.no_predecessors_iter()),
                               attr_dict=attr_dict)
+        # Insert the flow(s) retry if needed, and always make sure it
+        # is the **immediate** successor of the flow node itself.
         if flow.retry is not None:
             graph.add_node(flow.retry, kind=RETRY)
             _add_update_edges(graph, [flow], [flow.retry],
@@ -149,20 +184,42 @@ class FlowCompiler(object):
                 if node is not flow.retry and node is not flow:
                     graph.node[node].setdefault(RETRY, flow.retry)
             from_nodes = [flow.retry]
-            connected_attr_dict = {LINK_INVARIANT: True, LINK_RETRY: True}
+            attr_dict = {LINK_INVARIANT: True, LINK_RETRY: True}
         else:
             from_nodes = [flow]
-            connected_attr_dict = {LINK_INVARIANT: True}
-        connected_to = [
-            node for node in graph.no_predecessors_iter() if node is not flow
-        ]
-        if connected_to:
-            # Ensure all nodes in this graph(s) that have no
-            # predecessors depend on this flow (or this flow's retry) so that
-            # we can depend on the flow being traversed before its
-            # children (even though at the current time it will be skipped).
-            _add_update_edges(graph, from_nodes, connected_to,
-                              attr_dict=connected_attr_dict)
+            attr_dict = {LINK_INVARIANT: True}
+        # Ensure all nodes with no predecessors are connected to this flow
+        # or its retry node (so that the invariant that the flow node is
+        # traversed through before its contents is maintained); this allows
+        # us to easily know when we have entered a flow (when running) and
+        # do special and/or smart things such as only traverse up to the
+        # start of a flow when looking for node deciders.
+        _add_update_edges(graph, from_nodes, [
+            node for node in graph.no_predecessors_iter()
+            if node is not flow
+        ], attr_dict=attr_dict)
+        # Connect all nodes with no successors into a special terminator
+        # that is used to identify the end of the flow and ensure that all
+        # execution traversals will traverse over this node before executing
+        # further work (this is especially useful for nesting and knowing
+        # when we have exited a nesting level); it allows us to do special
+        # and/or smart things such as applying deciders up to (but not
+        # beyond) a flow termination point.
+        #
+        # Do note that in a empty flow this will just connect itself to
+        # the flow node itself... and also note we can not use the flow
+        # object itself (primarily because the underlying graph library
+        # uses hashing to identify node uniqueness and we can easily create
+        # a loop if we don't do this correctly, so avoid that by just
+        # creating this special node and tagging it with a special kind); we
+        # may be able to make this better in the future with a multidigraph
+        # that networkx provides??
+        flow_term = Terminator(flow)
+        graph.add_node(flow_term, kind=FLOW_END, noop=True)
+        _add_update_edges(graph, [
+            node for node in graph.no_successors_iter()
+            if node is not flow_term
+        ], [flow_term], attr_dict={LINK_INVARIANT: True})
         return graph, tree_node
 
 
diff --git a/taskflow/engines/action_engine/completer.py b/taskflow/engines/action_engine/completer.py
index d5c43a58b..02b63f761 100644
--- a/taskflow/engines/action_engine/completer.py
+++ b/taskflow/engines/action_engine/completer.py
@@ -128,7 +128,7 @@ class Completer(object):
         atom_states = self._storage.get_atoms_states(atom.name
                                                      for atom in atoms)
         for atom in atoms:
-            atom_state = atom_states[atom.name][0]
+            atom_state, _atom_intention = atom_states[atom.name]
             if atom_state == st.FAILURE:
                 self._process_atom_failure(atom, self._storage.get(atom.name))
         for retry in self._analyzer.iterate_retries(st.RETRYING):
@@ -137,7 +137,7 @@ class Completer(object):
                     atom_states[atom.name] = (state, intention)
         unfinished_atoms = set()
         for atom in atoms:
-            atom_state = atom_states[atom.name][0]
+            atom_state, _atom_intention = atom_states[atom.name]
             if atom_state in (st.RUNNING, st.REVERTING):
                 unfinished_atoms.add(atom)
         return unfinished_atoms
diff --git a/taskflow/engines/action_engine/deciders.py b/taskflow/engines/action_engine/deciders.py
new file mode 100644
index 000000000..4c4ac9d6c
--- /dev/null
+++ b/taskflow/engines/action_engine/deciders.py
@@ -0,0 +1,162 @@
+# -*- coding: utf-8 -*-
+
+#    Copyright (C) 2015 Yahoo! Inc. All Rights Reserved.
+#
+#    Licensed under the Apache License, Version 2.0 (the "License"); you may
+#    not use this file except in compliance with the License. You may obtain
+#    a copy of the License at
+#
+#         http://www.apache.org/licenses/LICENSE-2.0
+#
+#    Unless required by applicable law or agreed to in writing, software
+#    distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
+#    WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
+#    License for the specific language governing permissions and limitations
+#    under the License.
+
+import abc
+import itertools
+
+import six
+
+from taskflow import deciders
+from taskflow.engines.action_engine import compiler
+from taskflow.engines.action_engine import traversal
+from taskflow import states
+
+
+@six.add_metaclass(abc.ABCMeta)
+class Decider(object):
+    """Base class for deciders.
+
+    Provides interface to be implemented by sub-classes.
+
+    Deciders check whether next atom in flow should be executed or not.
+    """
+
+    @abc.abstractmethod
+    def tally(self, runtime):
+        """Tally edge deciders on whether this decider should allow running.
+
+        The returned value is a list of edge deciders that voted
+        'nay' (do not allow running).
+        """
+
+    @abc.abstractmethod
+    def affect(self, runtime, nay_voters):
+        """Affects associated atoms due to at least one 'nay' edge decider.
+
+        This will alter the associated atom + some set of successor atoms by
+        setting there state and intention to ``IGNORE`` so that they are
+        ignored in future runtime activities.
+        """
+
+    def check_and_affect(self, runtime):
+        """Handles :py:func:`~.tally` + :py:func:`~.affect` in right order.
+
+        NOTE(harlowja):  If there are zero 'nay' edge deciders then it is
+        assumed this decider should allow running.
+
+        Returns boolean of whether this decider allows for running (or not).
+        """
+        nay_voters = self.tally(runtime)
+        if nay_voters:
+            self.affect(runtime, nay_voters)
+            return False
+        return True
+
+
+def _affect_all_successors(atom, runtime):
+    execution_graph = runtime.compilation.execution_graph
+    successors_iter = traversal.depth_first_iterate(
+        execution_graph, atom, traversal.Direction.FORWARD)
+    runtime.reset_atoms(itertools.chain([atom], successors_iter),
+                        state=states.IGNORE, intention=states.IGNORE)
+
+
+def _affect_successor_tasks_in_same_flow(atom, runtime):
+    execution_graph = runtime.compilation.execution_graph
+    successors_iter = traversal.depth_first_iterate(
+        execution_graph, atom, traversal.Direction.FORWARD,
+        # Do not go through nested flows but do follow *all* tasks that
+        # are directly connected in this same flow (thus the reason this is
+        # called the same flow decider); retries are direct successors
+        # of flows, so they should also be not traversed through, but
+        # setting this explicitly ensures that.
+        through_flows=False, through_retries=False)
+    runtime.reset_atoms(itertools.chain([atom], successors_iter),
+                        state=states.IGNORE, intention=states.IGNORE)
+
+
+def _affect_atom(atom, runtime):
+    runtime.reset_atoms([atom], state=states.IGNORE, intention=states.IGNORE)
+
+
+def _affect_direct_task_neighbors(atom, runtime):
+    def _walk_neighbors():
+        execution_graph = runtime.compilation.execution_graph
+        for node in execution_graph.successors_iter(atom):
+            node_data = execution_graph.node[node]
+            if node_data['kind'] == compiler.TASK:
+                yield node
+    successors_iter = _walk_neighbors()
+    runtime.reset_atoms(itertools.chain([atom], successors_iter),
+                        state=states.IGNORE, intention=states.IGNORE)
+
+
+class IgnoreDecider(Decider):
+    """Checks any provided edge-deciders and determines if ok to run."""
+
+    _depth_strategies = {
+        deciders.Depth.ALL: _affect_all_successors,
+        deciders.Depth.ATOM: _affect_atom,
+        deciders.Depth.FLOW: _affect_successor_tasks_in_same_flow,
+        deciders.Depth.NEIGHBORS: _affect_direct_task_neighbors,
+    }
+
+    def __init__(self, atom, edge_deciders):
+        self._atom = atom
+        self._edge_deciders = edge_deciders
+
+    def tally(self, runtime):
+        if not self._edge_deciders:
+            return []
+        # Gather all atoms (the ones that were not ignored) results so that
+        # those results can be used
+        # by the decider(s) that are making a decision as to pass or
+        # not pass...
+        states_intentions = runtime.storage.get_atoms_states(
+            ed.from_node.name for ed in self._edge_deciders
+            if ed.kind in compiler.ATOMS)
+        history = {}
+        for atom_name in six.iterkeys(states_intentions):
+            atom_state, _atom_intention = states_intentions[atom_name]
+            if atom_state != states.IGNORE:
+                history[atom_name] = runtime.storage.get(atom_name)
+        nay_voters = []
+        for ed in self._edge_deciders:
+            if ed.kind in compiler.ATOMS and ed.from_node.name not in history:
+                continue
+            if not ed.decider(history=history):
+                nay_voters.append(ed)
+        return nay_voters
+
+    def affect(self, runtime, nay_voters):
+        # If there were many 'nay' edge deciders that were targeted
+        # at this atom, then we need to pick the one which has the widest
+        # impact and respect that one as the decider depth that will
+        # actually affect things.
+        widest_depth = deciders.pick_widest(ed.depth for ed in nay_voters)
+        affector = self._depth_strategies[widest_depth]
+        return affector(self._atom, runtime)
+
+
+class NoOpDecider(Decider):
+    """No-op decider that says it is always ok to run & has no effect(s)."""
+
+    def tally(self, runtime):
+        """Always good to go."""
+        return []
+
+    def affect(self, runtime, nay_voters):
+        """Does nothing."""
diff --git a/taskflow/engines/action_engine/engine.py b/taskflow/engines/action_engine/engine.py
index c65da3429..7084f51b9 100644
--- a/taskflow/engines/action_engine/engine.py
+++ b/taskflow/engines/action_engine/engine.py
@@ -99,37 +99,37 @@ class ActionEngine(base.Engine):
 
     **Engine options:**
 
-    +-------------------+-----------------------+------+-----------+
-    | Name/key          | Description           | Type | Default   |
-    +===================+=======================+======+===========+
-    | defer_reverts     | This option lets you  | bool | ``False`` |
-    |                   | safely nest flows     |      |           |
-    |                   | with retries inside   |      |           |
-    |                   | flows without retries |      |           |
-    |                   | and it still behaves  |      |           |
-    |                   | as a user would       |      |           |
-    |                   | expect (for example   |      |           |
-    |                   | if the retry gets     |      |           |
-    |                   | exhausted it reverts  |      |           |
-    |                   | the outer flow unless |      |           |
-    |                   | the outer flow has a  |      |           |
-    |                   | has a separate retry  |      |           |
-    |                   | behavior).            |      |           |
-    +-------------------+-----------------------+------+-----------+
-    | inject_transient  | When true, values     | bool | ``True``  |
-    |                   | that are local to     |      |           |
-    |                   | each atoms scope      |      |           |
-    |                   | are injected into     |      |           |
-    |                   | storage into a        |      |           |
-    |                   | transient location    |      |           |
-    |                   | (typically a local    |      |           |
-    |                   | dictionary), when     |      |           |
-    |                   | false those values    |      |           |
-    |                   | are instead persisted |      |           |
-    |                   | into atom details     |      |           |
-    |                   | (and saved in a non-  |      |           |
-    |                   | transient manner).    |      |           |
-    +-------------------+-----------------------+------+-----------+
+    +----------------------+-----------------------+------+------------+
+    | Name/key             | Description           | Type | Default    |
+    +======================+=======================+======+============+
+    | ``defer_reverts``    | This option lets you  | bool | ``False``  |
+    |                      | safely nest flows     |      |            |
+    |                      | with retries inside   |      |            |
+    |                      | flows without retries |      |            |
+    |                      | and it still behaves  |      |            |
+    |                      | as a user would       |      |            |
+    |                      | expect (for example   |      |            |
+    |                      | if the retry gets     |      |            |
+    |                      | exhausted it reverts  |      |            |
+    |                      | the outer flow unless |      |            |
+    |                      | the outer flow has a  |      |            |
+    |                      | has a separate retry  |      |            |
+    |                      | behavior).            |      |            |
+    +----------------------+-----------------------+------+------------+
+    | ``inject_transient`` | When true, values     | bool | ``True``   |
+    |                      | that are local to     |      |            |
+    |                      | each atoms scope      |      |            |
+    |                      | are injected into     |      |            |
+    |                      | storage into a        |      |            |
+    |                      | transient location    |      |            |
+    |                      | (typically a local    |      |            |
+    |                      | dictionary), when     |      |            |
+    |                      | false those values    |      |            |
+    |                      | are instead persisted |      |            |
+    |                      | into atom details     |      |            |
+    |                      | (and saved in a non-  |      |            |
+    |                      | transient manner).    |      |            |
+    +----------------------+-----------------------+------+------------+
     """
 
     NO_RERAISING_STATES = frozenset([states.SUSPENDED, states.SUCCESS])
@@ -148,6 +148,12 @@ class ActionEngine(base.Engine):
     end-users when doing execution iterations via :py:meth:`.run_iter`.
     """
 
+    MAX_MACHINE_STATES_RETAINED = 10
+    """
+    During :py:meth:`~.run_iter` the last X state machine transitions will
+    be recorded (typically only useful on failure).
+    """
+
     def __init__(self, flow, flow_detail, backend, options):
         super(ActionEngine, self).__init__(flow, flow_detail, backend, options)
         self._runtime = None
@@ -242,16 +248,21 @@ class ActionEngine(base.Engine):
         self.compile()
         self.prepare()
         self.validate()
-        last_state = None
+        # Keep track of the last X state changes, which if a failure happens
+        # are quite useful to log (and the performance of tracking this
+        # should be negligible).
+        last_transitions = collections.deque(
+            maxlen=max(1, self.MAX_MACHINE_STATES_RETAINED))
         with _start_stop(self._task_executor, self._retry_executor):
             self._change_state(states.RUNNING)
             try:
                 closed = False
                 machine, memory = self._runtime.builder.build(timeout=timeout)
                 r = runners.FiniteRunner(machine)
-                for (_prior_state, new_state) in r.run_iter(builder.START):
-                    last_state = new_state
-                    # NOTE(harlowja): skip over meta-states.
+                for transition in r.run_iter(builder.START):
+                    last_transitions.append(transition)
+                    _prior_state, new_state = transition
+                    # NOTE(harlowja): skip over meta-states
                     if new_state in builder.META_STATES:
                         continue
                     if new_state == states.FAILURE:
@@ -271,15 +282,24 @@ class ActionEngine(base.Engine):
                             self.suspend()
             except Exception:
                 with excutils.save_and_reraise_exception():
+                    LOG.exception("Engine execution has failed, something"
+                                  " bad must of happened (last"
+                                  " %s machine transitions were %s)",
+                                  last_transitions.maxlen,
+                                  list(last_transitions))
                     self._change_state(states.FAILURE)
             else:
-                if last_state and last_state not in self.IGNORABLE_STATES:
-                    self._change_state(new_state)
-                    if last_state not in self.NO_RERAISING_STATES:
-                        it = itertools.chain(
-                            six.itervalues(self.storage.get_failures()),
-                            six.itervalues(self.storage.get_revert_failures()))
-                        failure.Failure.reraise_if_any(it)
+                if last_transitions:
+                    _prior_state, new_state = last_transitions[-1]
+                    if new_state not in self.IGNORABLE_STATES:
+                        self._change_state(new_state)
+                        if new_state not in self.NO_RERAISING_STATES:
+                            failures = self.storage.get_failures()
+                            more_failures = self.storage.get_revert_failures()
+                            fails = itertools.chain(
+                                six.itervalues(failures),
+                                six.itervalues(more_failures))
+                            failure.Failure.reraise_if_any(fails)
 
     @staticmethod
     def _check_compilation(compilation):
diff --git a/taskflow/engines/action_engine/runtime.py b/taskflow/engines/action_engine/runtime.py
index 41dbd7798..f6fc9c15a 100644
--- a/taskflow/engines/action_engine/runtime.py
+++ b/taskflow/engines/action_engine/runtime.py
@@ -19,6 +19,7 @@ import functools
 
 from futurist import waiters
 
+from taskflow import deciders as de
 from taskflow.engines.action_engine.actions import retry as ra
 from taskflow.engines.action_engine.actions import task as ta
 from taskflow.engines.action_engine import analyzer as an
@@ -27,11 +28,17 @@ from taskflow.engines.action_engine import compiler as com
 from taskflow.engines.action_engine import completer as co
 from taskflow.engines.action_engine import scheduler as sched
 from taskflow.engines.action_engine import scopes as sc
+from taskflow.engines.action_engine import traversal as tr
 from taskflow import exceptions as exc
-from taskflow.flow import LINK_DECIDER
 from taskflow import states as st
 from taskflow.utils import misc
 
+from taskflow.flow import (LINK_DECIDER, LINK_DECIDER_DEPTH)  # noqa
+
+# Small helper to make the edge decider tuples more easily useable...
+_EdgeDecider = collections.namedtuple('_EdgeDecider',
+                                      'from_node,kind,decider,depth')
+
 
 class Runtime(object):
     """A aggregate of runtime objects, properties, ... used during execution.
@@ -42,7 +49,8 @@ class Runtime(object):
     """
 
     def __init__(self, compilation, storage, atom_notifier,
-                 task_executor, retry_executor, options=None):
+                 task_executor, retry_executor,
+                 options=None):
         self._atom_notifier = atom_notifier
         self._task_executor = task_executor
         self._retry_executor = retry_executor
@@ -51,11 +59,10 @@ class Runtime(object):
         self._atom_cache = {}
         self._options = misc.safe_copy_dict(options)
 
-    @staticmethod
-    def _walk_edge_deciders(graph, atom):
+    def _walk_edge_deciders(self, graph, atom):
         """Iterates through all nodes, deciders that alter atoms execution."""
         # This is basically a reverse breadth first exploration, with
-        # special logic to further traverse down flow nodes...
+        # special logic to further traverse down flow nodes as needed...
         predecessors_iter = graph.predecessors_iter
         nodes = collections.deque((u_node, atom)
                                   for u_node in predecessors_iter(atom))
@@ -63,14 +70,19 @@ class Runtime(object):
         while nodes:
             u_node, v_node = nodes.popleft()
             u_node_kind = graph.node[u_node]['kind']
+            u_v_data = graph.adj[u_node][v_node]
             try:
-                yield (u_node, u_node_kind,
-                       graph.adj[u_node][v_node][LINK_DECIDER])
+                decider = u_v_data[LINK_DECIDER]
+                decider_depth = u_v_data.get(LINK_DECIDER_DEPTH)
+                if decider_depth is None:
+                    decider_depth = de.Depth.ALL
+                yield _EdgeDecider(u_node, u_node_kind,
+                                   decider, decider_depth)
             except KeyError:
                 pass
             if u_node_kind == com.FLOW and u_node not in visited:
-                # Avoid re-exploring the same flow if we get to this
-                # same flow by a different *future* path...
+                # Avoid re-exploring the same flow if we get to this same
+                # flow by a different *future* path...
                 visited.add(u_node)
                 # Since we *currently* jump over flow node(s), we need to make
                 # sure that any prior decider that was directed at this flow
@@ -108,7 +120,7 @@ class Runtime(object):
         graph = self._compilation.execution_graph
         for node, node_data in graph.nodes_iter(data=True):
             node_kind = node_data['kind']
-            if node_kind == com.FLOW:
+            if node_kind in com.FLOWS:
                 continue
             elif node_kind in com.ATOMS:
                 check_transition_handler = check_transition_handlers[node_kind]
@@ -128,6 +140,10 @@ class Runtime(object):
             metadata['edge_deciders'] = tuple(deciders_it)
             metadata['action'] = action
             self._atom_cache[node.name] = metadata
+        # TODO(harlowja): optimize the different decider depths to avoid
+        # repeated full successor searching; this can be done by searching
+        # for the widest depth of parent(s), and limiting the search of
+        # children by the that depth.
 
     @property
     def compilation(self):
@@ -246,11 +262,12 @@ class Runtime(object):
     def reset_subgraph(self, atom, state=st.PENDING, intention=st.EXECUTE):
         """Resets a atoms subgraph to the given state and intention.
 
-        The subgraph is contained of all of the atoms successors.
+        The subgraph is contained of **all** of the atoms successors.
         """
-        return self.reset_atoms(
-            self.analyzer.iterate_connected_atoms(atom),
-            state=state, intention=intention)
+        execution_graph = self._compilation.execution_graph
+        atoms_it = tr.depth_first_iterate(execution_graph, atom,
+                                          tr.Direction.FORWARD)
+        return self.reset_atoms(atoms_it, state=state, intention=intention)
 
     def retry_subflow(self, retry):
         """Prepares a retrys + its subgraph for execution.
diff --git a/taskflow/engines/action_engine/scopes.py b/taskflow/engines/action_engine/scopes.py
index 4558ddda2..01a7546ec 100644
--- a/taskflow/engines/action_engine/scopes.py
+++ b/taskflow/engines/action_engine/scopes.py
@@ -15,33 +15,12 @@
 #    under the License.
 
 from taskflow.engines.action_engine import compiler as co
+from taskflow.engines.action_engine import traversal as tr
 from taskflow import logging
 
 LOG = logging.getLogger(__name__)
 
 
-def _depth_first_reverse_iterate(node, idx=-1):
-    """Iterates connected (in reverse) nodes (from starting node).
-
-    Jumps through nodes with ``FLOW`` ``kind`` attribute (does not yield
-    them back).
-    """
-    # Always go left to right, since right to left is the pattern order
-    # and we want to go backwards and not forwards through that ordering...
-    if idx == -1:
-        children_iter = node.reverse_iter()
-    else:
-        children_iter = reversed(node[0:idx])
-    for child in children_iter:
-        if child.metadata['kind'] == co.FLOW:
-            # Jump through these...
-            for child_child in child.dfs_iter(right_to_left=False):
-                if child_child.metadata['kind'] in co.ATOMS:
-                    yield child_child.item
-        else:
-            yield child.item
-
-
 class ScopeWalker(object):
     """Walks through the scopes of a atom using a engines compilation.
 
@@ -117,7 +96,9 @@ class ScopeWalker(object):
             except KeyError:
                 visible = []
                 removals = set()
-                for atom in _depth_first_reverse_iterate(parent, idx=last_idx):
+                atom_it = tr.depth_first_reverse_iterate(
+                    parent, start_from_idx=last_idx)
+                for atom in atom_it:
                     if atom in predecessors:
                         predecessors.remove(atom)
                         removals.add(atom)
diff --git a/taskflow/engines/action_engine/traversal.py b/taskflow/engines/action_engine/traversal.py
new file mode 100644
index 000000000..ad0b794ed
--- /dev/null
+++ b/taskflow/engines/action_engine/traversal.py
@@ -0,0 +1,126 @@
+# -*- coding: utf-8 -*-
+
+#    Copyright (C) 2015 Yahoo! Inc. All Rights Reserved.
+#
+#    Licensed under the Apache License, Version 2.0 (the "License"); you may
+#    not use this file except in compliance with the License. You may obtain
+#    a copy of the License at
+#
+#         http://www.apache.org/licenses/LICENSE-2.0
+#
+#    Unless required by applicable law or agreed to in writing, software
+#    distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
+#    WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
+#    License for the specific language governing permissions and limitations
+#    under the License.
+
+import collections
+import enum
+
+from taskflow.engines.action_engine import compiler as co
+
+
+class Direction(enum.Enum):
+    """Traversal direction enum."""
+
+    #: Go through successors.
+    FORWARD = 1
+
+    #: Go through predecessors.
+    BACKWARD = 2
+
+
+def _extract_connectors(execution_graph, starting_node, direction,
+                        through_flows=True, through_retries=True,
+                        through_tasks=True):
+    if direction == Direction.FORWARD:
+        connected_iter = execution_graph.successors_iter
+    else:
+        connected_iter = execution_graph.predecessors_iter
+    connected_to_functors = {}
+    if through_flows:
+        connected_to_functors[co.FLOW] = connected_iter
+        connected_to_functors[co.FLOW_END] = connected_iter
+    if through_retries:
+        connected_to_functors[co.RETRY] = connected_iter
+    if through_tasks:
+        connected_to_functors[co.TASK] = connected_iter
+    return connected_iter(starting_node), connected_to_functors
+
+
+def breadth_first_iterate(execution_graph, starting_node, direction,
+                          through_flows=True, through_retries=True,
+                          through_tasks=True):
+    """Iterates connected nodes in execution graph (from starting node).
+
+    Does so in a breadth first manner.
+
+    Jumps over nodes with ``noop`` attribute (does not yield them back).
+    """
+    initial_nodes_iter, connected_to_functors = _extract_connectors(
+        execution_graph, starting_node, direction,
+        through_flows=through_flows, through_retries=through_retries,
+        through_tasks=through_tasks)
+    q = collections.deque(initial_nodes_iter)
+    while q:
+        node = q.popleft()
+        node_attrs = execution_graph.node[node]
+        if not node_attrs.get('noop'):
+            yield node
+        try:
+            node_kind = node_attrs['kind']
+            connected_to_functor = connected_to_functors[node_kind]
+        except KeyError:
+            pass
+        else:
+            q.extend(connected_to_functor(node))
+
+
+def depth_first_iterate(execution_graph, starting_node, direction,
+                        through_flows=True, through_retries=True,
+                        through_tasks=True):
+    """Iterates connected nodes in execution graph (from starting node).
+
+    Does so in a depth first manner.
+
+    Jumps over nodes with ``noop`` attribute (does not yield them back).
+    """
+    initial_nodes_iter, connected_to_functors = _extract_connectors(
+        execution_graph, starting_node, direction,
+        through_flows=through_flows, through_retries=through_retries,
+        through_tasks=through_tasks)
+    stack = list(initial_nodes_iter)
+    while stack:
+        node = stack.pop()
+        node_attrs = execution_graph.node[node]
+        if not node_attrs.get('noop'):
+            yield node
+        try:
+            node_kind = node_attrs['kind']
+            connected_to_functor = connected_to_functors[node_kind]
+        except KeyError:
+            pass
+        else:
+            stack.extend(connected_to_functor(node))
+
+
+def depth_first_reverse_iterate(node, start_from_idx=-1):
+    """Iterates connected (in reverse) **tree** nodes (from starting node).
+
+    Jumps through nodes with ``noop`` attribute (does not yield them back).
+    """
+    # Always go left to right, since right to left is the pattern order
+    # and we want to go backwards and not forwards through that ordering...
+    if start_from_idx == -1:
+        # All of them...
+        children_iter = node.reverse_iter()
+    else:
+        children_iter = reversed(node[0:start_from_idx])
+    for child in children_iter:
+        if child.metadata.get('noop'):
+            # Jump through these...
+            for grand_child in child.dfs_iter(right_to_left=False):
+                if grand_child.metadata['kind'] in co.ATOMS:
+                    yield grand_child.item
+        else:
+            yield child.item
diff --git a/taskflow/exceptions.py b/taskflow/exceptions.py
index 857d6634d..beb66e999 100644
--- a/taskflow/exceptions.py
+++ b/taskflow/exceptions.py
@@ -221,6 +221,14 @@ class InvalidFormat(TaskFlowException):
     """Raised when some object/entity is not in the expected format."""
 
 
+class DisallowedAccess(TaskFlowException):
+    """Raised when storage access is not possible due to state limitations."""
+
+    def __init__(self, message, cause=None, state=None):
+        super(DisallowedAccess, self).__init__(message, cause=cause)
+        self.state = state
+
+
 # Others.
 
 class NotImplementedError(NotImplementedError):
diff --git a/taskflow/flow.py b/taskflow/flow.py
index a0f6846bd..3b974f7b0 100644
--- a/taskflow/flow.py
+++ b/taskflow/flow.py
@@ -39,6 +39,9 @@ LINK_DECIDER = 'decider'
 _CHOP_PAT = "taskflow.patterns."
 _CHOP_PAT_LEN = len(_CHOP_PAT)
 
+# This key denotes the depth the decider will apply (defaulting to all).
+LINK_DECIDER_DEPTH = 'decider_depth'
+
 
 @six.add_metaclass(abc.ABCMeta)
 class Flow(object):
diff --git a/taskflow/formatters.py b/taskflow/formatters.py
index 41b409c98..af6c6a50f 100644
--- a/taskflow/formatters.py
+++ b/taskflow/formatters.py
@@ -42,16 +42,20 @@ def _fetch_predecessor_tree(graph, atom):
     """Creates a tree of predecessors, rooted at given atom."""
     root = tree.Node(atom)
     stack = [(root, atom)]
-    seen = set()
     while stack:
         parent, node = stack.pop()
         for pred_node in graph.predecessors_iter(node):
-            child = tree.Node(pred_node,
-                              **graph.node[pred_node])
-            parent.add(child)
-            stack.append((child, pred_node))
-            seen.add(pred_node)
-    return len(seen), root
+            pred_node_data = graph.node[pred_node]
+            if pred_node_data['kind'] == compiler.FLOW_END:
+                # Jump over and/or don't show flow end nodes...
+                for pred_pred_node in graph.predecessors_iter(pred_node):
+                    stack.append((parent, pred_pred_node))
+            else:
+                child = tree.Node(pred_node, **pred_node_data)
+                parent.add(child)
+                # And go further backwards...
+                stack.append((child, pred_node))
+    return root
 
 
 class FailureFormatter(object):
@@ -64,59 +68,51 @@ class FailureFormatter(object):
     def __init__(self, engine, hide_inputs_outputs_of=()):
         self._hide_inputs_outputs_of = hide_inputs_outputs_of
         self._engine = engine
-        self._formatter_funcs = {
-            compiler.FLOW: self._format_flow,
-        }
-        for kind in compiler.ATOMS:
-            self._formatter_funcs[kind] = self._format_atom
-
-    def _format_atom(self, storage, cache, node):
-        """Formats a single tree node (atom) into a string version."""
-        atom = node.item
-        atom_name = atom.name
-        atom_attrs = {}
-        intention, intention_found = _cached_get(cache, 'intentions',
-                                                 atom_name,
-                                                 storage.get_atom_intention,
-                                                 atom_name)
-        if intention_found:
-            atom_attrs['intention'] = intention
-        state, state_found = _cached_get(cache, 'states', atom_name,
-                                         storage.get_atom_state, atom_name)
-        if state_found:
-            atom_attrs['state'] = state
-        if atom_name not in self._hide_inputs_outputs_of:
-            # When the cache does not exist for this atom this
-            # will be called with the rest of these arguments
-            # used to populate the cache.
-            fetch_mapped_args = functools.partial(
-                storage.fetch_mapped_args, atom.rebind,
-                atom_name=atom_name, optional_args=atom.optional)
-            requires, requires_found = _cached_get(cache, 'requires',
-                                                   atom_name,
-                                                   fetch_mapped_args)
-            if requires_found:
-                atom_attrs['requires'] = requires
-            provides, provides_found = _cached_get(cache, 'provides',
-                                                   atom_name,
-                                                   storage.get_execute_result,
-                                                   atom_name)
-            if provides_found:
-                atom_attrs['provides'] = provides
-        if atom_attrs:
-            return "Atom '%s' %s" % (atom_name, atom_attrs)
-        else:
-            return "Atom '%s'" % (atom_name)
-
-    def _format_flow(self, storage, cache, node):
-        """Formats a single tree node (flow) into a string version."""
-        flow = node.item
-        return flow.name
 
     def _format_node(self, storage, cache, node):
         """Formats a single tree node into a string version."""
-        formatter_func = self. _formatter_funcs[node.metadata['kind']]
-        return formatter_func(storage, cache, node)
+        if node.metadata['kind'] == compiler.FLOW:
+            flow = node.item
+            flow_name = flow.name
+            return "Flow '%s'" % (flow_name)
+        elif node.metadata['kind'] in compiler.ATOMS:
+            atom = node.item
+            atom_name = atom.name
+            atom_attrs = {}
+            intention, intention_found = _cached_get(
+                cache, 'intentions', atom_name, storage.get_atom_intention,
+                atom_name)
+            if intention_found:
+                atom_attrs['intention'] = intention
+            state, state_found = _cached_get(cache, 'states', atom_name,
+                                             storage.get_atom_state,
+                                             atom_name)
+            if state_found:
+                atom_attrs['state'] = state
+            if atom_name not in self._hide_inputs_outputs_of:
+                # When the cache does not exist for this atom this
+                # will be called with the rest of these arguments
+                # used to populate the cache.
+                fetch_mapped_args = functools.partial(
+                    storage.fetch_mapped_args, atom.rebind,
+                    atom_name=atom_name, optional_args=atom.optional)
+                requires, requires_found = _cached_get(cache, 'requires',
+                                                       atom_name,
+                                                       fetch_mapped_args)
+                if requires_found:
+                    atom_attrs['requires'] = requires
+                provides, provides_found = _cached_get(
+                    cache, 'provides', atom_name,
+                    storage.get_execute_result, atom_name)
+                if provides_found:
+                    atom_attrs['provides'] = provides
+            if atom_attrs:
+                return "Atom '%s' %s" % (atom_name, atom_attrs)
+            else:
+                return "Atom '%s'" % (atom_name)
+        else:
+            raise TypeError("Unable to format node, unknown node"
+                            " kind '%s' encountered" % node.metadata['kind'])
 
     def format(self, fail, atom_matcher):
         """Returns a (exc_info, details) tuple about the failure.
@@ -143,13 +139,11 @@ class FailureFormatter(object):
         graph = compilation.execution_graph
         atom_node = hierarchy.find_first_match(atom_matcher)
         atom = None
-        priors = 0
         atom_intention = None
         if atom_node is not None:
             atom = atom_node.item
             atom_intention = storage.get_atom_intention(atom.name)
-            priors = sum(c for (_n, c) in graph.in_degree_iter([atom]))
-        if atom is not None and priors and atom_intention in self._BUILDERS:
+        if atom is not None and atom_intention in self._BUILDERS:
             # Cache as much as we can, since the path of various atoms
             # may cause the same atom to be seen repeatedly depending on
             # the graph structure...
@@ -160,12 +154,13 @@ class FailureFormatter(object):
                 'states': {},
             }
             builder, kind = self._BUILDERS[atom_intention]
-            count, rooted_tree = builder(graph, atom)
-            buff.write_nl('%s %s (most recent atoms first):' % (count, kind))
+            rooted_tree = builder(graph, atom)
+            child_count = rooted_tree.child_count(only_direct=False)
+            buff.write_nl('%s %s (most recent first):' % (child_count, kind))
             formatter = functools.partial(self._format_node, storage, cache)
-            child_count = rooted_tree.child_count()
+            direct_child_count = rooted_tree.child_count(only_direct=True)
             for i, child in enumerate(rooted_tree, 1):
-                if i == child_count:
+                if i == direct_child_count:
                     buff.write(child.pformat(stringify_node=formatter,
                                              starting_prefix="  "))
                 else:
diff --git a/taskflow/patterns/graph_flow.py b/taskflow/patterns/graph_flow.py
index c769124f0..e3d9d9b26 100644
--- a/taskflow/patterns/graph_flow.py
+++ b/taskflow/patterns/graph_flow.py
@@ -18,6 +18,7 @@ import collections
 
 import six
 
+from taskflow import deciders as de
 from taskflow import exceptions as exc
 from taskflow import flow
 from taskflow.types import graph as gr
@@ -73,7 +74,7 @@ class Flow(flow.Flow):
     #: Extracts the unsatisified symbol requirements of a single node.
     _unsatisfied_requires = staticmethod(_unsatisfied_requires)
 
-    def link(self, u, v, decider=None):
+    def link(self, u, v, decider=None, decider_depth=None):
         """Link existing node u as a runtime dependency of existing node v.
 
         Note that if the addition of these edges creates a `cyclic`_ graph
@@ -93,6 +94,13 @@ class Flow(flow.Flow):
                         links that have ``v`` as a target. It is expected to
                         return a single boolean (``True`` to allow ``v``
                         execution or ``False`` to not).
+        :param decider_depth: One of the :py:class:`~taskflow.deciders.Depth`
+                              enumerations (or a string version of) that will
+                              be used to influence what atoms are ignored
+                              when the decider provided results false. If
+                              not provided (and a valid decider is provided
+                              then this defaults to
+                              :py:attr:`~taskflow.deciders.Depth.ALL`).
 
         .. _cyclic: https://en.wikipedia.org/wiki/Cycle_graph
         """
@@ -103,11 +111,13 @@ class Flow(flow.Flow):
         if decider is not None:
             if not six.callable(decider):
                 raise ValueError("Decider boolean callback must be callable")
-        self._swap(self._link(u, v, manual=True, decider=decider))
+        self._swap(self._link(u, v, manual=True,
+                              decider=decider, decider_depth=decider_depth))
         return self
 
     def _link(self, u, v, graph=None,
-              reason=None, manual=False, decider=None):
+              reason=None, manual=False, decider=None,
+              decider_depth=None):
         mutable_graph = True
         if graph is None:
             graph = self._graph
@@ -119,6 +129,18 @@ class Flow(flow.Flow):
             attrs = {}
         if decider is not None:
             attrs[flow.LINK_DECIDER] = decider
+            try:
+                # Remove existing decider depth, if one existed.
+                del attrs[flow.LINK_DECIDER_DEPTH]
+            except KeyError:
+                pass
+        if decider_depth is not None:
+            if decider is None:
+                raise ValueError("Decider depth requires a decider to be"
+                                 " provided along with it")
+            else:
+                decider_depth = de.Depth.translate(decider_depth)
+                attrs[flow.LINK_DECIDER_DEPTH] = decider_depth
         if manual:
             attrs[flow.LINK_MANUAL] = True
         if reason is not None:
diff --git a/taskflow/storage.py b/taskflow/storage.py
index 1784ff749..5c348f350 100644
--- a/taskflow/storage.py
+++ b/taskflow/storage.py
@@ -15,6 +15,7 @@
 #    under the License.
 
 import contextlib
+import functools
 
 import fasteners
 from oslo_utils import reflection
@@ -84,6 +85,127 @@ META_PROGRESS = 'progress'
 META_PROGRESS_DETAILS = 'progress_details'
 
 
+class _ProviderLocator(object):
+    """Helper to start to better decouple the finding logic from storage.
+
+    WIP: part of the larger effort to cleanup/refactor the finding of named
+         arguments so that the code can be more unified and easy to
+         follow...
+    """
+
+    def __init__(self, transient_results,
+                 providers_fetcher, result_fetcher):
+        self.result_fetcher = result_fetcher
+        self.providers_fetcher = providers_fetcher
+        self.transient_results = transient_results
+
+    def _try_get_results(self, looking_for, provider,
+                         look_into_results=True, find_potentials=False):
+        if provider.name is _TRANSIENT_PROVIDER:
+            # TODO(harlowja): This 'is' check still sucks, do this
+            # better in the future...
+            results = self.transient_results
+        else:
+            try:
+                results = self.result_fetcher(provider.name)
+            except (exceptions.NotFound, exceptions.DisallowedAccess):
+                if not find_potentials:
+                    raise
+                else:
+                    # Ok, likely hasn't produced a result yet, but
+                    # at a future point it hopefully will, so stub
+                    # out the *expected* result.
+                    results = {}
+        if look_into_results:
+            _item_from_single(provider, results, looking_for)
+        return results
+
+    def _find(self, looking_for, scope_walker=None,
+              short_circuit=True, find_potentials=False):
+        if scope_walker is None:
+            scope_walker = []
+        default_providers, atom_providers = self.providers_fetcher(looking_for)
+        searched_providers = set()
+        providers_and_results = []
+        if default_providers:
+            for p in default_providers:
+                searched_providers.add(p)
+                try:
+                    provider_results = self._try_get_results(
+                        looking_for, p, find_potentials=find_potentials,
+                        # For default providers always look into there
+                        # results as default providers are statically setup
+                        # and therefore looking into there provided results
+                        # should fail early.
+                        look_into_results=True)
+                except exceptions.NotFound:
+                    if not find_potentials:
+                        raise
+                else:
+                    providers_and_results.append((p, provider_results))
+            if short_circuit:
+                return (searched_providers, providers_and_results)
+        if not atom_providers:
+            return (searched_providers, providers_and_results)
+        atom_providers_by_name = dict((p.name, p) for p in atom_providers)
+        for accessible_atom_names in iter(scope_walker):
+            # *Always* retain the scope ordering (if any matches
+            # happen); instead of retaining the possible provider match
+            # order (which isn't that important and may be different from
+            # the scope requested ordering).
+            maybe_atom_providers = [atom_providers_by_name[atom_name]
+                                    for atom_name in accessible_atom_names
+                                    if atom_name in atom_providers_by_name]
+            tmp_providers_and_results = []
+            if find_potentials:
+                for p in maybe_atom_providers:
+                    searched_providers.add(p)
+                    tmp_providers_and_results.append((p, {}))
+            else:
+                for p in maybe_atom_providers:
+                    searched_providers.add(p)
+                    try:
+                        # Don't at this point look into the provider results
+                        # as calling code will grab all providers, and then
+                        # get the result from the *first* provider that
+                        # actually provided it (or die).
+                        provider_results = self._try_get_results(
+                            looking_for, p, find_potentials=find_potentials,
+                            look_into_results=False)
+                    except exceptions.DisallowedAccess as e:
+                        if e.state != states.IGNORE:
+                            exceptions.raise_with_cause(
+                                exceptions.NotFound,
+                                "Expected to be able to find output %r"
+                                " produced by %s but was unable to get at"
+                                " that providers results" % (looking_for, p))
+                        else:
+                            LOG.blather("Avoiding using the results of"
+                                        " %r (from %s) for name %r because"
+                                        " it was ignored", p.name, p,
+                                        looking_for)
+                    else:
+                        tmp_providers_and_results.append((p, provider_results))
+            if tmp_providers_and_results and short_circuit:
+                return (searched_providers, tmp_providers_and_results)
+            else:
+                providers_and_results.extend(tmp_providers_and_results)
+        return (searched_providers, providers_and_results)
+
+    def find_potentials(self, looking_for, scope_walker=None):
+        """Returns the accessible **potential** providers."""
+        _searched_providers, providers_and_results = self._find(
+            looking_for, scope_walker=scope_walker,
+            short_circuit=False, find_potentials=True)
+        return set(p for (p, _provider_results) in providers_and_results)
+
+    def find(self, looking_for, scope_walker=None, short_circuit=True):
+        """Returns the accessible providers."""
+        return self._find(looking_for, scope_walker=scope_walker,
+                          short_circuit=short_circuit,
+                          find_potentials=False)
+
+
 class _Provider(object):
     """A named symbol provider that produces a output at the given index."""
 
@@ -326,13 +448,13 @@ class Storage(object):
             ad = self._flowdetail.find(self._atom_name_to_uuid[atom_name])
         except KeyError:
             exceptions.raise_with_cause(exceptions.NotFound,
-                                        "Unknown atom name: %s" % atom_name)
+                                        "Unknown atom name '%s'" % atom_name)
         else:
             # TODO(harlowja): we need to figure out how to get away from doing
             # these kinds of type checks in general (since they likely mean
             # we aren't doing something right).
             if expected_type and not isinstance(ad, expected_type):
-                raise TypeError("Atom %s is not of the expected type: %s"
+                raise TypeError("Atom '%s' is not of the expected type: %s"
                                 % (atom_name,
                                    reflection.get_class_name(expected_type)))
             if clone:
@@ -479,8 +601,9 @@ class Storage(object):
             try:
                 _item_from(container, index)
             except _EXTRACTION_EXCEPTIONS:
-                LOG.warning("Atom %s did not supply result "
-                            "with index %r (name %s)", atom_name, index, name)
+                LOG.warning("Atom '%s' did not supply result "
+                            "with index %r (name '%s')", atom_name, index,
+                            name)
 
     @fasteners.write_locked
     def save(self, atom_name, result, state=states.SUCCESS):
@@ -545,17 +668,34 @@ class Storage(object):
             except KeyError:
                 pass
             return failure
-        # TODO(harlowja): this seems like it should be checked before fetching
-        # the potential failure, instead of after, fix this soon...
-        if source.state not in allowed_states:
-            raise exceptions.NotFound("Result for atom %s is not currently"
-                                      " known" % atom_name)
-        return getattr(source, results_attr_name)
+        else:
+            if source.state not in allowed_states:
+                raise exceptions.DisallowedAccess(
+                    "Result for atom '%s' is not known/accessible"
+                    " due to it being in %s state when result access"
+                    " is restricted to %s states" % (atom_name,
+                                                     source.state,
+                                                     allowed_states),
+                    state=source.state)
+            return getattr(source, results_attr_name)
 
     def get_execute_result(self, atom_name):
         """Gets the ``execute`` results for an atom from storage."""
-        return self._get(atom_name, 'results', 'failure',
-                         _EXECUTE_STATES_WITH_RESULTS, states.EXECUTE)
+        try:
+            results = self._get(atom_name, 'results', 'failure',
+                                _EXECUTE_STATES_WITH_RESULTS, states.EXECUTE)
+        except exceptions.DisallowedAccess as e:
+            if e.state == states.IGNORE:
+                exceptions.raise_with_cause(exceptions.NotFound,
+                                            "Result for atom '%s' execution"
+                                            " is not known (as it was"
+                                            " ignored)" % atom_name)
+            else:
+                exceptions.raise_with_cause(exceptions.NotFound,
+                                            "Result for atom '%s' execution"
+                                            " is not known" % atom_name)
+        else:
+            return results
 
     @fasteners.read_locked
     def _get_failures(self, fail_cache_key):
@@ -577,8 +717,21 @@ class Storage(object):
 
     def get_revert_result(self, atom_name):
         """Gets the ``revert`` results for an atom from storage."""
-        return self._get(atom_name, 'revert_results', 'revert_failure',
-                         _REVERT_STATES_WITH_RESULTS, states.REVERT)
+        try:
+            results = self._get(atom_name, 'revert_results', 'revert_failure',
+                                _REVERT_STATES_WITH_RESULTS, states.REVERT)
+        except exceptions.DisallowedAccess as e:
+            if e.state == states.IGNORE:
+                exceptions.raise_with_cause(exceptions.NotFound,
+                                            "Result for atom '%s' revert is"
+                                            " not known (as it was"
+                                            " ignored)" % atom_name)
+            else:
+                exceptions.raise_with_cause(exceptions.NotFound,
+                                            "Result for atom '%s' revert is"
+                                            " not known" % atom_name)
+        else:
+            return results
 
     def get_revert_failures(self):
         """Get all ``revert`` failures that happened with this flow."""
@@ -639,7 +792,7 @@ class Storage(object):
             be serializable).
         """
         if atom_name not in self._atom_name_to_uuid:
-            raise exceptions.NotFound("Unknown atom name: %s" % atom_name)
+            raise exceptions.NotFound("Unknown atom name '%s'" % atom_name)
 
         def save_transient():
             self._injected_args.setdefault(atom_name, {})
@@ -728,6 +881,19 @@ class Storage(object):
         self._set_result_mapping(provider_name,
                                  dict((name, name) for name in names))
 
+    def _fetch_providers(self, looking_for, providers=None):
+        """Return pair of (default providers, atom providers)."""
+        if providers is None:
+            providers = self._reverse_mapping.get(looking_for, [])
+        default_providers = []
+        atom_providers = []
+        for p in providers:
+            if p.name in (_TRANSIENT_PROVIDER, self.injector_name):
+                default_providers.append(p)
+            else:
+                atom_providers.append(p)
+        return default_providers, atom_providers
+
     def _set_result_mapping(self, provider_name, mapping):
         """Sets the result mapping for a given producer.
 
@@ -757,30 +923,30 @@ class Storage(object):
         if many_handler is None:
             many_handler = _many_handler
         try:
-            providers = self._reverse_mapping[name]
+            maybe_providers = self._reverse_mapping[name]
         except KeyError:
-            exceptions.raise_with_cause(exceptions.NotFound,
-                                        "Name %r is not mapped as a produced"
-                                        " output by any providers" % name)
+            raise exceptions.NotFound("Name %r is not mapped as a produced"
+                                      " output by any providers" % name)
+        locator = _ProviderLocator(
+            self._transients,
+            functools.partial(self._fetch_providers,
+                              providers=maybe_providers),
+            lambda atom_name:
+                self._get(atom_name, 'last_results', 'failure',
+                          _EXECUTE_STATES_WITH_RESULTS, states.EXECUTE))
         values = []
-        for provider in providers:
-            if provider.name is _TRANSIENT_PROVIDER:
-                values.append(_item_from_single(provider,
-                                                self._transients, name))
-            else:
-                try:
-                    container = self._get(provider.name,
-                                          'last_results', 'failure',
-                                          _EXECUTE_STATES_WITH_RESULTS,
-                                          states.EXECUTE)
-                except exceptions.NotFound:
-                    pass
-                else:
-                    values.append(_item_from_single(provider,
-                                                    container, name))
+        searched_providers, providers = locator.find(
+            name, short_circuit=False,
+            # NOTE(harlowja): There are no scopes used here (as of now), so
+            # we just return all known providers as if it was one large
+            # scope.
+            scope_walker=[[p.name for p in maybe_providers]])
+        for provider, results in providers:
+            values.append(_item_from_single(provider, results, name))
         if not values:
-            raise exceptions.NotFound("Unable to find result %r,"
-                                      " searched %s" % (name, providers))
+            raise exceptions.NotFound(
+                "Unable to find result %r, searched %s providers"
+                % (name, len(searched_providers)))
         else:
             return many_handler(values)
 
@@ -796,49 +962,6 @@ class Storage(object):
         needed values; it just checks that they are registered to produce
         it in the future.
         """
-
-        def _fetch_providers(name):
-            """Fetchs pair of (default providers, non-default providers)."""
-            default_providers = []
-            non_default_providers = []
-            for p in self._reverse_mapping.get(name, []):
-                if p.name in (_TRANSIENT_PROVIDER, self.injector_name):
-                    default_providers.append(p)
-                else:
-                    non_default_providers.append(p)
-            return default_providers, non_default_providers
-
-        def _locate_providers(name, scope_walker=None):
-            """Finds the accessible *potential* providers."""
-            default_providers, non_default_providers = _fetch_providers(name)
-            providers = []
-            if non_default_providers:
-                if scope_walker is not None:
-                    scope_iter = iter(scope_walker)
-                else:
-                    scope_iter = iter([])
-                for names in scope_iter:
-                    for p in non_default_providers:
-                        if p.name in names:
-                            providers.append(p)
-            for p in default_providers:
-                if p.name is _TRANSIENT_PROVIDER:
-                    results = self._transients
-                else:
-                    try:
-                        results = self._get(p.name, 'last_results', 'failure',
-                                            _EXECUTE_STATES_WITH_RESULTS,
-                                            states.EXECUTE)
-                    except exceptions.NotFound:
-                        results = {}
-                try:
-                    _item_from_single(p, results, name)
-                except exceptions.NotFound:
-                    pass
-                else:
-                    providers.append(p)
-            return providers
-
         source, _clone = self._atomdetail_by_name(atom_name)
         if scope_walker is None:
             scope_walker = self._scope_fetcher(atom_name)
@@ -849,6 +972,11 @@ class Storage(object):
             source.meta.get(META_INJECTED, {}),
         ]
         missing = set(six.iterkeys(args_mapping))
+        locator = _ProviderLocator(
+            self._transients, self._fetch_providers,
+            lambda atom_name:
+                self._get(atom_name, 'last_results', 'failure',
+                          _EXECUTE_STATES_WITH_RESULTS, states.EXECUTE))
         for (bound_name, name) in six.iteritems(args_mapping):
             if LOG.isEnabledFor(logging.TRACE):
                 LOG.trace("Looking for %r <= %r for atom '%s'",
@@ -863,8 +991,8 @@ class Storage(object):
                     continue
                 if name in source:
                     maybe_providers += 1
-            providers = _locate_providers(name, scope_walker=scope_walker)
-            maybe_providers += len(providers)
+            maybe_providers += len(
+                locator.find_potentials(name, scope_walker=scope_walker))
             if maybe_providers:
                 LOG.trace("Atom '%s' will have %s potential providers"
                           " of %r <= %r", atom_name, maybe_providers,
@@ -894,7 +1022,6 @@ class Storage(object):
                           atom_name=None, scope_walker=None,
                           optional_args=None):
         """Fetch ``execute`` arguments for an atom using its args mapping."""
-
         def _extract_first_from(name, sources):
             """Extracts/returns first occurence of key in list of dicts."""
             for i, source in enumerate(sources):
@@ -903,49 +1030,6 @@ class Storage(object):
                 if name in source:
                     return (i, source[name])
             raise KeyError(name)
-
-        def _get_results(looking_for, provider):
-            """Gets the results saved for a given provider."""
-            try:
-                return self._get(provider.name, 'last_results', 'failure',
-                                 _EXECUTE_STATES_WITH_RESULTS,
-                                 states.EXECUTE)
-            except exceptions.NotFound:
-                exceptions.raise_with_cause(exceptions.NotFound,
-                                            "Expected to be able to find"
-                                            " output %r produced by %s but was"
-                                            " unable to get at that providers"
-                                            " results" % (looking_for,
-                                                          provider))
-
-        def _locate_providers(looking_for, possible_providers,
-                              scope_walker=None):
-            """Finds the accessible providers."""
-            default_providers = []
-            for p in possible_providers:
-                if p.name is _TRANSIENT_PROVIDER:
-                    default_providers.append((p, self._transients))
-                if p.name == self.injector_name:
-                    default_providers.append((p, _get_results(looking_for, p)))
-            if default_providers:
-                return default_providers
-            if scope_walker is not None:
-                scope_iter = iter(scope_walker)
-            else:
-                scope_iter = iter([])
-            extractor = lambda p: p.name
-            for names in scope_iter:
-                # *Always* retain the scope ordering (if any matches
-                # happen); instead of retaining the possible provider match
-                # order (which isn't that important and may be different from
-                # the scope requested ordering).
-                providers = misc.look_for(names, possible_providers,
-                                          extractor=extractor)
-                if providers:
-                    return [(p, _get_results(looking_for, p))
-                            for p in providers]
-            return []
-
         if optional_args is None:
             optional_args = []
         if atom_name:
@@ -960,6 +1044,9 @@ class Storage(object):
             injected_sources = []
         if not args_mapping:
             return {}
+        get_results = lambda atom_name: \
+            self._get(atom_name, 'last_results', 'failure',
+                      _EXECUTE_STATES_WITH_RESULTS, states.EXECUTE)
         mapped_args = {}
         for (bound_name, name) in six.iteritems(args_mapping):
             if LOG.isEnabledFor(logging.TRACE):
@@ -969,8 +1056,8 @@ class Storage(object):
                 else:
                     LOG.trace("Looking for %r <= %r", bound_name, name)
             try:
-                source_index, value = _extract_first_from(name,
-                                                          injected_sources)
+                source_index, value = _extract_first_from(
+                    name, injected_sources)
                 mapped_args[bound_name] = value
                 if LOG.isEnabledFor(logging.TRACE):
                     if source_index == 0:
@@ -983,7 +1070,7 @@ class Storage(object):
                                   " values)", bound_name, name, value)
             except KeyError:
                 try:
-                    possible_providers = self._reverse_mapping[name]
+                    maybe_providers = self._reverse_mapping[name]
                 except KeyError:
                     if bound_name in optional_args:
                         LOG.trace("Argument %r is optional, skipping",
@@ -992,15 +1079,18 @@ class Storage(object):
                     raise exceptions.NotFound("Name %r is not mapped as a"
                                               " produced output by any"
                                               " providers" % name)
-                # Reduce the possible providers to one that are allowed.
-                providers = _locate_providers(name, possible_providers,
-                                              scope_walker=scope_walker)
+                locator = _ProviderLocator(
+                    self._transients,
+                    functools.partial(self._fetch_providers,
+                                      providers=maybe_providers), get_results)
+                searched_providers, providers = locator.find(
+                    name, scope_walker=scope_walker)
                 if not providers:
                     raise exceptions.NotFound(
                         "Mapped argument %r <= %r was not produced"
                         " by any accessible provider (%s possible"
                         " providers were scanned)"
-                        % (bound_name, name, len(possible_providers)))
+                        % (bound_name, name, len(searched_providers)))
                 provider, value = _item_from_first_of(providers, name)
                 mapped_args[bound_name] = value
                 LOG.trace("Matched %r <= %r to %r (from %s)",
diff --git a/taskflow/tests/unit/action_engine/test_compile.py b/taskflow/tests/unit/action_engine/test_compile.py
index 6ccf35889..757bde78d 100644
--- a/taskflow/tests/unit/action_engine/test_compile.py
+++ b/taskflow/tests/unit/action_engine/test_compile.py
@@ -25,12 +25,27 @@ from taskflow import test
 from taskflow.tests import utils as test_utils
 
 
+def _replicate_graph_with_names(compilation):
+    # Turn a graph of nodes into a graph of names only so that
+    # testing can use those names instead of having to use the exact
+    # node objects themselves (which is problematic for any end nodes that
+    # are added into the graph *dynamically*, and are not there in the
+    # original/source flow).
+    g = compilation.execution_graph
+    n_g = g.__class__(name=g.name)
+    for node, node_data in g.nodes_iter(data=True):
+        n_g.add_node(node.name, attr_dict=node_data)
+    for u, v, u_v_data in g.edges_iter(data=True):
+        n_g.add_edge(u.name, v.name, attr_dict=u_v_data)
+    return n_g
+
+
 class PatternCompileTest(test.TestCase):
     def test_task(self):
         task = test_utils.DummyTask(name='a')
-        compilation = compiler.PatternCompiler(task).compile()
-        g = compilation.execution_graph
-        self.assertEqual([task], list(g.nodes()))
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(task).compile())
+        self.assertEqual(['a'], list(g.nodes()))
         self.assertEqual([], list(g.edges()))
 
     def test_retry(self):
@@ -54,19 +69,20 @@ class PatternCompileTest(test.TestCase):
         inner_flo.add(d)
         flo.add(inner_flo)
 
-        compilation = compiler.PatternCompiler(flo).compile()
-        g = compilation.execution_graph
-        self.assertEqual(6, len(g))
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
+        self.assertEqual(8, len(g))
 
         order = g.topological_sort()
-        self.assertEqual([flo, a, b, c, inner_flo, d], order)
-        self.assertTrue(g.has_edge(c, inner_flo))
-        self.assertTrue(g.has_edge(inner_flo, d))
+        self.assertEqual(['test', 'a', 'b', 'c',
+                          "sub-test", 'd', "sub-test[$]",
+                          'test[$]'], order)
+        self.assertTrue(g.has_edge('c', "sub-test"))
+        self.assertTrue(g.has_edge("sub-test", 'd'))
         self.assertEqual({'invariant': True},
-                         g.get_edge_data(inner_flo, d))
-
-        self.assertEqual([d], list(g.no_successors_iter()))
-        self.assertEqual([flo], list(g.no_predecessors_iter()))
+                         g.get_edge_data("sub-test", 'd'))
+        self.assertEqual(['test[$]'], list(g.no_successors_iter()))
+        self.assertEqual(['test'], list(g.no_predecessors_iter()))
 
     def test_invalid(self):
         a, b, c = test_utils.make_many(3)
@@ -80,19 +96,21 @@ class PatternCompileTest(test.TestCase):
         a, b, c, d = test_utils.make_many(4)
         flo = uf.Flow("test")
         flo.add(a, b, c, d)
-        compilation = compiler.PatternCompiler(flo).compile()
-        g = compilation.execution_graph
-        self.assertEqual(5, len(g))
+
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
+        self.assertEqual(6, len(g))
         self.assertItemsEqual(g.edges(), [
-            (flo, a),
-            (flo, b),
-            (flo, c),
-            (flo, d),
+            ('test', 'a'),
+            ('test', 'b'),
+            ('test', 'c'),
+            ('test', 'd'),
+            ('a', 'test[$]'),
+            ('b', 'test[$]'),
+            ('c', 'test[$]'),
+            ('d', 'test[$]'),
         ])
-        self.assertEqual(set([a, b, c, d]),
-                         set(g.no_successors_iter()))
-        self.assertEqual(set([flo]),
-                         set(g.no_predecessors_iter()))
+        self.assertEqual(set(['test']), set(g.no_predecessors_iter()))
 
     def test_linear_nested(self):
         a, b, c, d = test_utils.make_many(4)
@@ -102,22 +120,22 @@ class PatternCompileTest(test.TestCase):
         inner_flo.add(c, d)
         flo.add(inner_flo)
 
-        compilation = compiler.PatternCompiler(flo).compile()
-        graph = compilation.execution_graph
-        self.assertEqual(6, len(graph))
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
+        self.assertEqual(8, len(g))
 
-        lb = graph.subgraph([a, b])
-        self.assertFalse(lb.has_edge(b, a))
-        self.assertTrue(lb.has_edge(a, b))
-        self.assertEqual({'invariant': True}, graph.get_edge_data(a, b))
+        sub_g = g.subgraph(['a', 'b'])
+        self.assertFalse(sub_g.has_edge('b', 'a'))
+        self.assertTrue(sub_g.has_edge('a', 'b'))
+        self.assertEqual({'invariant': True}, sub_g.get_edge_data("a", "b"))
 
-        ub = graph.subgraph([c, d])
-        self.assertEqual(0, ub.number_of_edges())
+        sub_g = g.subgraph(['c', 'd'])
+        self.assertEqual(0, sub_g.number_of_edges())
 
         # This ensures that c and d do not start executing until after b.
-        self.assertTrue(graph.has_edge(b, inner_flo))
-        self.assertTrue(graph.has_edge(inner_flo, c))
-        self.assertTrue(graph.has_edge(inner_flo, d))
+        self.assertTrue(g.has_edge('b', 'test2'))
+        self.assertTrue(g.has_edge('test2', 'c'))
+        self.assertTrue(g.has_edge('test2', 'd'))
 
     def test_unordered_nested(self):
         a, b, c, d = test_utils.make_many(4)
@@ -127,15 +145,19 @@ class PatternCompileTest(test.TestCase):
         flo2.add(c, d)
         flo.add(flo2)
 
-        compilation = compiler.PatternCompiler(flo).compile()
-        g = compilation.execution_graph
-        self.assertEqual(6, len(g))
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
+        self.assertEqual(8, len(g))
         self.assertItemsEqual(g.edges(), [
-            (flo, a),
-            (flo, b),
-            (flo, flo2),
-            (flo2, c),
-            (c, d)
+            ('test', 'a'),
+            ('test', 'b'),
+            ('test', 'test2'),
+            ('test2', 'c'),
+            ('c', 'd'),
+            ('d', 'test2[$]'),
+            ('test2[$]', 'test[$]'),
+            ('a', 'test[$]'),
+            ('b', 'test[$]'),
         ])
 
     def test_unordered_nested_in_linear(self):
@@ -143,27 +165,27 @@ class PatternCompileTest(test.TestCase):
         inner_flo = uf.Flow('ut').add(b, c)
         flo = lf.Flow('lt').add(a, inner_flo, d)
 
-        compilation = compiler.PatternCompiler(flo).compile()
-        g = compilation.execution_graph
-        self.assertEqual(6, len(g))
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
+        self.assertEqual(8, len(g))
         self.assertItemsEqual(g.edges(), [
-            (flo, a),
-            (a, inner_flo),
-            (inner_flo, b),
-            (inner_flo, c),
-            (b, d),
-            (c, d),
+            ('lt', 'a'),
+            ('a', 'ut'),
+            ('ut', 'b'),
+            ('ut', 'c'),
+            ('b', 'ut[$]'),
+            ('c', 'ut[$]'),
+            ('ut[$]', 'd'),
+            ('d', 'lt[$]'),
         ])
 
     def test_graph(self):
         a, b, c, d = test_utils.make_many(4)
         flo = gf.Flow("test")
         flo.add(a, b, c, d)
-
         compilation = compiler.PatternCompiler(flo).compile()
-        g = compilation.execution_graph
-        self.assertEqual(5, len(g))
-        self.assertEqual(4, g.number_of_edges())
+        self.assertEqual(6, len(compilation.execution_graph))
+        self.assertEqual(8, compilation.execution_graph.number_of_edges())
 
     def test_graph_nested(self):
         a, b, c, d, e, f, g = test_utils.make_many(7)
@@ -174,19 +196,26 @@ class PatternCompileTest(test.TestCase):
         flo2.add(e, f, g)
         flo.add(flo2)
 
-        compilation = compiler.PatternCompiler(flo).compile()
-        graph = compilation.execution_graph
-        self.assertEqual(9, len(graph))
-        self.assertItemsEqual(graph.edges(), [
-            (flo, a),
-            (flo, b),
-            (flo, c),
-            (flo, d),
-            (flo, flo2),
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
+        self.assertEqual(11, len(g))
+        self.assertItemsEqual(g.edges(), [
+            ('test', 'a'),
+            ('test', 'b'),
+            ('test', 'c'),
+            ('test', 'd'),
+            ('a', 'test[$]'),
+            ('b', 'test[$]'),
+            ('c', 'test[$]'),
+            ('d', 'test[$]'),
 
-            (flo2, e),
-            (e, f),
-            (f, g),
+            ('test', 'test2'),
+            ('test2', 'e'),
+            ('e', 'f'),
+            ('f', 'g'),
+
+            ('g', 'test2[$]'),
+            ('test2[$]', 'test[$]'),
         ])
 
     def test_graph_nested_graph(self):
@@ -198,19 +227,29 @@ class PatternCompileTest(test.TestCase):
         flo2.add(e, f, g)
         flo.add(flo2)
 
-        compilation = compiler.PatternCompiler(flo).compile()
-        graph = compilation.execution_graph
-        self.assertEqual(9, len(graph))
-        self.assertItemsEqual(graph.edges(), [
-            (flo, a),
-            (flo, b),
-            (flo, c),
-            (flo, d),
-            (flo, flo2),
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
+        self.assertEqual(11, len(g))
+        self.assertItemsEqual(g.edges(), [
+            ('test', 'a'),
+            ('test', 'b'),
+            ('test', 'c'),
+            ('test', 'd'),
+            ('test', 'test2'),
 
-            (flo2, e),
-            (flo2, f),
-            (flo2, g),
+            ('test2', 'e'),
+            ('test2', 'f'),
+            ('test2', 'g'),
+
+            ('e', 'test2[$]'),
+            ('f', 'test2[$]'),
+            ('g', 'test2[$]'),
+
+            ('test2[$]', 'test[$]'),
+            ('a', 'test[$]'),
+            ('b', 'test[$]'),
+            ('c', 'test[$]'),
+            ('d', 'test[$]'),
         ])
 
     def test_graph_links(self):
@@ -221,33 +260,34 @@ class PatternCompileTest(test.TestCase):
         flo.link(b, c)
         flo.link(c, d)
 
-        compilation = compiler.PatternCompiler(flo).compile()
-        g = compilation.execution_graph
-        self.assertEqual(5, len(g))
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
+        self.assertEqual(6, len(g))
         self.assertItemsEqual(g.edges(data=True), [
-            (flo, a, {'invariant': True}),
-
-            (a, b, {'manual': True}),
-            (b, c, {'manual': True}),
-            (c, d, {'manual': True}),
+            ('test', 'a', {'invariant': True}),
+            ('a', 'b', {'manual': True}),
+            ('b', 'c', {'manual': True}),
+            ('c', 'd', {'manual': True}),
+            ('d', 'test[$]', {'invariant': True}),
         ])
-        self.assertItemsEqual([flo], g.no_predecessors_iter())
-        self.assertItemsEqual([d], g.no_successors_iter())
+        self.assertItemsEqual(['test'], g.no_predecessors_iter())
+        self.assertItemsEqual(['test[$]'], g.no_successors_iter())
 
     def test_graph_dependencies(self):
         a = test_utils.ProvidesRequiresTask('a', provides=['x'], requires=[])
         b = test_utils.ProvidesRequiresTask('b', provides=[], requires=['x'])
         flo = gf.Flow("test").add(a, b)
 
-        compilation = compiler.PatternCompiler(flo).compile()
-        g = compilation.execution_graph
-        self.assertEqual(3, len(g))
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
+        self.assertEqual(4, len(g))
         self.assertItemsEqual(g.edges(data=True), [
-            (flo, a, {'invariant': True}),
-            (a, b, {'reasons': set(['x'])})
+            ('test', 'a', {'invariant': True}),
+            ('a', 'b', {'reasons': set(['x'])}),
+            ('b', 'test[$]', {'invariant': True}),
         ])
-        self.assertItemsEqual([flo], g.no_predecessors_iter())
-        self.assertItemsEqual([b], g.no_successors_iter())
+        self.assertItemsEqual(['test'], g.no_predecessors_iter())
+        self.assertItemsEqual(['test[$]'], g.no_successors_iter())
 
     def test_graph_nested_requires(self):
         a = test_utils.ProvidesRequiresTask('a', provides=['x'], requires=[])
@@ -256,17 +296,19 @@ class PatternCompileTest(test.TestCase):
         inner_flo = lf.Flow("test2").add(b, c)
         flo = gf.Flow("test").add(a, inner_flo)
 
-        compilation = compiler.PatternCompiler(flo).compile()
-        graph = compilation.execution_graph
-        self.assertEqual(5, len(graph))
-        self.assertItemsEqual(graph.edges(data=True), [
-            (flo, a, {'invariant': True}),
-            (inner_flo, b, {'invariant': True}),
-            (a, inner_flo, {'reasons': set(['x'])}),
-            (b, c, {'invariant': True}),
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
+        self.assertEqual(7, len(g))
+        self.assertItemsEqual(g.edges(data=True), [
+            ('test', 'a', {'invariant': True}),
+            ('test2', 'b', {'invariant': True}),
+            ('a', 'test2', {'reasons': set(['x'])}),
+            ('b', 'c', {'invariant': True}),
+            ('c', 'test2[$]', {'invariant': True}),
+            ('test2[$]', 'test[$]', {'invariant': True}),
         ])
-        self.assertItemsEqual([flo], graph.no_predecessors_iter())
-        self.assertItemsEqual([c], graph.no_successors_iter())
+        self.assertItemsEqual(['test'], list(g.no_predecessors_iter()))
+        self.assertItemsEqual(['test[$]'], list(g.no_successors_iter()))
 
     def test_graph_nested_provides(self):
         a = test_utils.ProvidesRequiresTask('a', provides=[], requires=['x'])
@@ -275,18 +317,22 @@ class PatternCompileTest(test.TestCase):
         inner_flo = lf.Flow("test2").add(b, c)
         flo = gf.Flow("test").add(a, inner_flo)
 
-        compilation = compiler.PatternCompiler(flo).compile()
-        graph = compilation.execution_graph
-        self.assertEqual(5, len(graph))
-        self.assertItemsEqual(graph.edges(data=True), [
-            (flo, inner_flo, {'invariant': True}),
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
+        self.assertEqual(7, len(g))
+        self.assertItemsEqual(g.edges(data=True), [
+            ('test', 'test2', {'invariant': True}),
+            ('a', 'test[$]', {'invariant': True}),
 
-            (inner_flo, b, {'invariant': True}),
-            (b, c, {'invariant': True}),
-            (c, a, {'reasons': set(['x'])}),
+            # The 'x' requirement is produced out of test2...
+            ('test2[$]', 'a', {'reasons': set(['x'])}),
+
+            ('test2', 'b', {'invariant': True}),
+            ('b', 'c', {'invariant': True}),
+            ('c', 'test2[$]', {'invariant': True}),
         ])
-        self.assertItemsEqual([flo], graph.no_predecessors_iter())
-        self.assertItemsEqual([a], graph.no_successors_iter())
+        self.assertItemsEqual(['test'], g.no_predecessors_iter())
+        self.assertItemsEqual(['test[$]'], g.no_successors_iter())
 
     def test_empty_flow_in_linear_flow(self):
         flo = lf.Flow('lf')
@@ -295,12 +341,14 @@ class PatternCompileTest(test.TestCase):
         empty_flo = gf.Flow("empty")
         flo.add(a, empty_flo, b)
 
-        compilation = compiler.PatternCompiler(flo).compile()
-        graph = compilation.execution_graph
-        self.assertItemsEqual(graph.edges(), [
-            (flo, a),
-            (a, empty_flo),
-            (empty_flo, b),
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
+        self.assertItemsEqual(g.edges(), [
+            ("lf", "a"),
+            ("a", "empty"),
+            ("empty", "empty[$]"),
+            ("empty[$]", "b"),
+            ("b", "lf[$]"),
         ])
 
     def test_many_empty_in_graph_flow(self):
@@ -331,22 +379,24 @@ class PatternCompileTest(test.TestCase):
         flo.link(a, d)
         flo.link(c, d)
 
-        compilation = compiler.PatternCompiler(flo).compile()
-        graph = compilation.execution_graph
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
 
-        self.assertTrue(graph.has_edge(flo, a))
+        self.assertTrue(g.has_edge('root', 'a'))
+        self.assertTrue(g.has_edge('root', 'b'))
+        self.assertTrue(g.has_edge('root', 'c'))
 
-        self.assertTrue(graph.has_edge(flo, b))
-        self.assertTrue(graph.has_edge(b_0, b_1))
-        self.assertTrue(graph.has_edge(b_1, b_2))
-        self.assertTrue(graph.has_edge(b_2, b_3))
+        self.assertTrue(g.has_edge('b.0', 'b.1'))
+        self.assertTrue(g.has_edge('b.1[$]', 'b.2'))
+        self.assertTrue(g.has_edge('b.2[$]', 'b.3'))
 
-        self.assertTrue(graph.has_edge(flo, c))
-        self.assertTrue(graph.has_edge(c_0, c_1))
-        self.assertTrue(graph.has_edge(c_1, c_2))
+        self.assertTrue(g.has_edge('c.0[$]', 'c.1'))
+        self.assertTrue(g.has_edge('c.1[$]', 'c.2'))
 
-        self.assertTrue(graph.has_edge(b_3, d))
-        self.assertEqual(12, len(graph))
+        self.assertTrue(g.has_edge('a', 'd'))
+        self.assertTrue(g.has_edge('b[$]', 'd'))
+        self.assertTrue(g.has_edge('c[$]', 'd'))
+        self.assertEqual(20, len(g))
 
     def test_empty_flow_in_nested_flow(self):
         flow = lf.Flow('lf')
@@ -360,13 +410,13 @@ class PatternCompileTest(test.TestCase):
         flow2.add(c, empty_flow, d)
         flow.add(a, flow2, b)
 
-        compilation = compiler.PatternCompiler(flow).compile()
-        g = compilation.execution_graph
-
-        for source, target in [(flow, a), (a, flow2),
-                               (flow2, c), (c, empty_flow),
-                               (empty_flow, d), (d, b)]:
-            self.assertTrue(g.has_edge(source, target))
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flow).compile())
+        for u, v in [('lf', 'a'), ('a', 'lf-2'),
+                     ('lf-2', 'c'), ('c', 'empty'),
+                     ('empty[$]', 'd'), ('d', 'lf-2[$]'),
+                     ('lf-2[$]', 'b'), ('b', 'lf[$]')]:
+            self.assertTrue(g.has_edge(u, v))
 
     def test_empty_flow_in_graph_flow(self):
         flow = lf.Flow('lf')
@@ -379,7 +429,14 @@ class PatternCompileTest(test.TestCase):
         g = compilation.execution_graph
         self.assertTrue(g.has_edge(flow, a))
         self.assertTrue(g.has_edge(a, empty_flow))
-        self.assertTrue(g.has_edge(empty_flow, b))
+
+        empty_flow_successors = g.successors(empty_flow)
+        self.assertEqual(1, len(empty_flow_successors))
+        empty_flow_terminal = empty_flow_successors[0]
+        self.assertIs(empty_flow, empty_flow_terminal.flow)
+        self.assertEqual(compiler.FLOW_END,
+                         g.node[empty_flow_terminal]['kind'])
+        self.assertTrue(g.has_edge(empty_flow_terminal, b))
 
     def test_empty_flow_in_graph_flow_linkage(self):
         flow = gf.Flow('lf')
@@ -417,146 +474,154 @@ class PatternCompileTest(test.TestCase):
     def test_retry_in_linear_flow(self):
         flo = lf.Flow("test", retry.AlwaysRevert("c"))
         compilation = compiler.PatternCompiler(flo).compile()
-        g = compilation.execution_graph
-        self.assertEqual(2, len(g))
-        self.assertEqual(1, g.number_of_edges())
+        self.assertEqual(3, len(compilation.execution_graph))
+        self.assertEqual(2, compilation.execution_graph.number_of_edges())
 
     def test_retry_in_unordered_flow(self):
         flo = uf.Flow("test", retry.AlwaysRevert("c"))
         compilation = compiler.PatternCompiler(flo).compile()
-        g = compilation.execution_graph
-        self.assertEqual(2, len(g))
-        self.assertEqual(1, g.number_of_edges())
+        self.assertEqual(3, len(compilation.execution_graph))
+        self.assertEqual(2, compilation.execution_graph.number_of_edges())
 
     def test_retry_in_graph_flow(self):
         flo = gf.Flow("test", retry.AlwaysRevert("c"))
         compilation = compiler.PatternCompiler(flo).compile()
         g = compilation.execution_graph
-        self.assertEqual(2, len(g))
-        self.assertEqual(1, g.number_of_edges())
+        self.assertEqual(3, len(g))
+        self.assertEqual(2, g.number_of_edges())
 
     def test_retry_in_nested_flows(self):
         c1 = retry.AlwaysRevert("c1")
         c2 = retry.AlwaysRevert("c2")
         inner_flo = lf.Flow("test2", c2)
         flo = lf.Flow("test", c1).add(inner_flo)
-        compilation = compiler.PatternCompiler(flo).compile()
-        g = compilation.execution_graph
 
-        self.assertEqual(4, len(g))
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
+        self.assertEqual(6, len(g))
         self.assertItemsEqual(g.edges(data=True), [
-            (flo, c1, {'invariant': True}),
-            (c1, inner_flo, {'invariant': True, 'retry': True}),
-            (inner_flo, c2, {'invariant': True}),
+            ('test', 'c1', {'invariant': True}),
+            ('c1', 'test2', {'invariant': True, 'retry': True}),
+            ('test2', 'c2', {'invariant': True}),
+            ('c2', 'test2[$]', {'invariant': True}),
+            ('test2[$]', 'test[$]', {'invariant': True}),
         ])
-        self.assertIs(c1, g.node[c2]['retry'])
-        self.assertItemsEqual([flo], g.no_predecessors_iter())
-        self.assertItemsEqual([c2], g.no_successors_iter())
+        self.assertIs(c1, g.node['c2']['retry'])
+        self.assertItemsEqual(['test'], list(g.no_predecessors_iter()))
+        self.assertItemsEqual(['test[$]'], list(g.no_successors_iter()))
 
     def test_retry_in_linear_flow_with_tasks(self):
         c = retry.AlwaysRevert("c")
         a, b = test_utils.make_many(2)
         flo = lf.Flow("test", c).add(a, b)
-        compilation = compiler.PatternCompiler(flo).compile()
-        g = compilation.execution_graph
 
-        self.assertEqual(4, len(g))
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
+        self.assertEqual(5, len(g))
         self.assertItemsEqual(g.edges(data=True), [
-            (flo, c, {'invariant': True}),
-            (a, b, {'invariant': True}),
-            (c, a, {'invariant': True, 'retry': True})
+            ('test', 'c', {'invariant': True}),
+            ('a', 'b', {'invariant': True}),
+            ('c', 'a', {'invariant': True, 'retry': True}),
+            ('b', 'test[$]', {'invariant': True}),
         ])
 
-        self.assertItemsEqual([flo], g.no_predecessors_iter())
-        self.assertItemsEqual([b], g.no_successors_iter())
-        self.assertIs(c, g.node[a]['retry'])
-        self.assertIs(c, g.node[b]['retry'])
+        self.assertItemsEqual(['test'], g.no_predecessors_iter())
+        self.assertItemsEqual(['test[$]'], g.no_successors_iter())
+        self.assertIs(c, g.node['a']['retry'])
+        self.assertIs(c, g.node['b']['retry'])
 
     def test_retry_in_unordered_flow_with_tasks(self):
         c = retry.AlwaysRevert("c")
         a, b = test_utils.make_many(2)
         flo = uf.Flow("test", c).add(a, b)
-        compilation = compiler.PatternCompiler(flo).compile()
-        g = compilation.execution_graph
 
-        self.assertEqual(4, len(g))
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
+        self.assertEqual(5, len(g))
         self.assertItemsEqual(g.edges(data=True), [
-            (flo, c, {'invariant': True}),
-            (c, a, {'invariant': True, 'retry': True}),
-            (c, b, {'invariant': True, 'retry': True}),
+            ('test', 'c', {'invariant': True}),
+            ('c', 'a', {'invariant': True, 'retry': True}),
+            ('c', 'b', {'invariant': True, 'retry': True}),
+            ('b', 'test[$]', {'invariant': True}),
+            ('a', 'test[$]', {'invariant': True}),
         ])
 
-        self.assertItemsEqual([flo], g.no_predecessors_iter())
-        self.assertItemsEqual([a, b], g.no_successors_iter())
-        self.assertIs(c, g.node[a]['retry'])
-        self.assertIs(c, g.node[b]['retry'])
+        self.assertItemsEqual(['test'], list(g.no_predecessors_iter()))
+        self.assertItemsEqual(['test[$]'], list(g.no_successors_iter()))
+        self.assertIs(c, g.node['a']['retry'])
+        self.assertIs(c, g.node['b']['retry'])
 
     def test_retry_in_graph_flow_with_tasks(self):
-        r = retry.AlwaysRevert("cp")
+        r = retry.AlwaysRevert("r")
         a, b, c = test_utils.make_many(3)
         flo = gf.Flow("test", r).add(a, b, c).link(b, c)
-        compilation = compiler.PatternCompiler(flo).compile()
-        g = compilation.execution_graph
-        self.assertEqual(5, len(g))
 
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
         self.assertItemsEqual(g.edges(data=True), [
-            (flo, r, {'invariant': True}),
-            (r, a, {'invariant': True, 'retry': True}),
-            (r, b, {'invariant': True, 'retry': True}),
-            (b, c, {'manual': True})
+            ('test', 'r', {'invariant': True}),
+            ('r', 'a', {'invariant': True, 'retry': True}),
+            ('r', 'b', {'invariant': True, 'retry': True}),
+            ('b', 'c', {'manual': True}),
+            ('a', 'test[$]', {'invariant': True}),
+            ('c', 'test[$]', {'invariant': True}),
         ])
 
-        self.assertItemsEqual([flo], g.no_predecessors_iter())
-        self.assertItemsEqual([a, c], g.no_successors_iter())
-        self.assertIs(r, g.node[a]['retry'])
-        self.assertIs(r, g.node[b]['retry'])
-        self.assertIs(r, g.node[c]['retry'])
+        self.assertItemsEqual(['test'], g.no_predecessors_iter())
+        self.assertItemsEqual(['test[$]'], g.no_successors_iter())
+        self.assertIs(r, g.node['a']['retry'])
+        self.assertIs(r, g.node['b']['retry'])
+        self.assertIs(r, g.node['c']['retry'])
 
     def test_retries_hierarchy(self):
-        c1 = retry.AlwaysRevert("cp1")
-        c2 = retry.AlwaysRevert("cp2")
+        c1 = retry.AlwaysRevert("c1")
+        c2 = retry.AlwaysRevert("c2")
         a, b, c, d = test_utils.make_many(4)
-        inner_flo = lf.Flow("test", c2).add(b, c)
+        inner_flo = lf.Flow("test2", c2).add(b, c)
         flo = lf.Flow("test", c1).add(a, inner_flo, d)
-        compilation = compiler.PatternCompiler(flo).compile()
-        g = compilation.execution_graph
 
-        self.assertEqual(8, len(g))
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
+        self.assertEqual(10, len(g))
         self.assertItemsEqual(g.edges(data=True), [
-            (flo, c1, {'invariant': True}),
-            (c1, a, {'invariant': True, 'retry': True}),
-            (a, inner_flo, {'invariant': True}),
-            (inner_flo, c2, {'invariant': True}),
-            (c2, b, {'invariant': True, 'retry': True}),
-            (b, c, {'invariant': True}),
-            (c, d, {'invariant': True}),
+            ('test', 'c1', {'invariant': True}),
+            ('c1', 'a', {'invariant': True, 'retry': True}),
+            ('a', 'test2', {'invariant': True}),
+            ('test2', 'c2', {'invariant': True}),
+            ('c2', 'b', {'invariant': True, 'retry': True}),
+            ('b', 'c', {'invariant': True}),
+            ('c', 'test2[$]', {'invariant': True}),
+            ('test2[$]', 'd', {'invariant': True}),
+            ('d', 'test[$]', {'invariant': True}),
         ])
-        self.assertIs(c1, g.node[a]['retry'])
-        self.assertIs(c1, g.node[d]['retry'])
-        self.assertIs(c2, g.node[b]['retry'])
-        self.assertIs(c2, g.node[c]['retry'])
-        self.assertIs(c1, g.node[c2]['retry'])
-        self.assertIs(None, g.node[c1].get('retry'))
+        self.assertIs(c1, g.node['a']['retry'])
+        self.assertIs(c1, g.node['d']['retry'])
+        self.assertIs(c2, g.node['b']['retry'])
+        self.assertIs(c2, g.node['c']['retry'])
+        self.assertIs(c1, g.node['c2']['retry'])
+        self.assertIs(None, g.node['c1'].get('retry'))
 
     def test_retry_subflows_hierarchy(self):
-        c1 = retry.AlwaysRevert("cp1")
+        c1 = retry.AlwaysRevert("c1")
         a, b, c, d = test_utils.make_many(4)
-        inner_flo = lf.Flow("test").add(b, c)
+        inner_flo = lf.Flow("test2").add(b, c)
         flo = lf.Flow("test", c1).add(a, inner_flo, d)
-        compilation = compiler.PatternCompiler(flo).compile()
-        g = compilation.execution_graph
 
-        self.assertEqual(7, len(g))
+        g = _replicate_graph_with_names(
+            compiler.PatternCompiler(flo).compile())
+        self.assertEqual(9, len(g))
         self.assertItemsEqual(g.edges(data=True), [
-            (flo, c1, {'invariant': True}),
-            (c1, a, {'invariant': True, 'retry': True}),
-            (a, inner_flo, {'invariant': True}),
-            (inner_flo, b, {'invariant': True}),
-            (b, c, {'invariant': True}),
-            (c, d, {'invariant': True}),
+            ('test', 'c1', {'invariant': True}),
+            ('c1', 'a', {'invariant': True, 'retry': True}),
+            ('a', 'test2', {'invariant': True}),
+            ('test2', 'b', {'invariant': True}),
+            ('b', 'c', {'invariant': True}),
+            ('c', 'test2[$]', {'invariant': True}),
+            ('test2[$]', 'd', {'invariant': True}),
+            ('d', 'test[$]', {'invariant': True}),
         ])
-        self.assertIs(c1, g.node[a]['retry'])
-        self.assertIs(c1, g.node[d]['retry'])
-        self.assertIs(c1, g.node[b]['retry'])
-        self.assertIs(c1, g.node[c]['retry'])
-        self.assertIs(None, g.node[c1].get('retry'))
+        self.assertIs(c1, g.node['a']['retry'])
+        self.assertIs(c1, g.node['d']['retry'])
+        self.assertIs(c1, g.node['b']['retry'])
+        self.assertIs(c1, g.node['c']['retry'])
+        self.assertIs(None, g.node['c1'].get('retry'))
diff --git a/taskflow/tests/unit/patterns/test_graph_flow.py b/taskflow/tests/unit/patterns/test_graph_flow.py
index 7197dc721..429f68fdb 100644
--- a/taskflow/tests/unit/patterns/test_graph_flow.py
+++ b/taskflow/tests/unit/patterns/test_graph_flow.py
@@ -26,6 +26,16 @@ def _task(name, provides=None, requires=None):
 
 
 class GraphFlowTest(test.TestCase):
+    def test_invalid_decider_depth(self):
+        g_1 = utils.ProgressingTask(name='g-1')
+        g_2 = utils.ProgressingTask(name='g-2')
+        for not_a_depth in ['not-a-depth', object(), 2, 3.4, False]:
+            flow = gf.Flow('g')
+            flow.add(g_1, g_2)
+            self.assertRaises((ValueError, TypeError),
+                              flow.link, g_1, g_2,
+                              decider=lambda history: False,
+                              decider_depth=not_a_depth)
 
     def test_graph_flow_stringy(self):
         f = gf.Flow('test')
diff --git a/taskflow/tests/unit/test_deciders.py b/taskflow/tests/unit/test_deciders.py
new file mode 100644
index 000000000..8bfc154d6
--- /dev/null
+++ b/taskflow/tests/unit/test_deciders.py
@@ -0,0 +1,58 @@
+# -*- coding: utf-8 -*-
+
+#    Copyright (C) 2015 Yahoo! Inc. All Rights Reserved.
+#
+#    Licensed under the Apache License, Version 2.0 (the "License"); you may
+#    not use this file except in compliance with the License. You may obtain
+#    a copy of the License at
+#
+#         http://www.apache.org/licenses/LICENSE-2.0
+#
+#    Unless required by applicable law or agreed to in writing, software
+#    distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
+#    WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
+#    License for the specific language governing permissions and limitations
+#    under the License.
+
+from taskflow import deciders
+from taskflow import test
+
+
+class TestDeciders(test.TestCase):
+    def test_translate(self):
+        for val in ['all', 'ALL', 'aLL', deciders.Depth.ALL]:
+            self.assertEqual(deciders.Depth.ALL,
+                             deciders.Depth.translate(val))
+        for val in ['atom', 'ATOM', 'atOM', deciders.Depth.ATOM]:
+            self.assertEqual(deciders.Depth.ATOM,
+                             deciders.Depth.translate(val))
+        for val in ['neighbors', 'Neighbors',
+                    'NEIGHBORS', deciders.Depth.NEIGHBORS]:
+            self.assertEqual(deciders.Depth.NEIGHBORS,
+                             deciders.Depth.translate(val))
+        for val in ['flow', 'FLOW', 'flOW', deciders.Depth.FLOW]:
+            self.assertEqual(deciders.Depth.FLOW,
+                             deciders.Depth.translate(val))
+
+    def test_bad_translate(self):
+        self.assertRaises(TypeError, deciders.Depth.translate, 3)
+        self.assertRaises(TypeError, deciders.Depth.translate, object())
+        self.assertRaises(ValueError, deciders.Depth.translate, "stuff")
+
+    def test_pick_widest(self):
+        choices = [deciders.Depth.ATOM, deciders.Depth.FLOW]
+        self.assertEqual(deciders.Depth.FLOW, deciders.pick_widest(choices))
+        choices = [deciders.Depth.ATOM, deciders.Depth.FLOW,
+                   deciders.Depth.ALL]
+        self.assertEqual(deciders.Depth.ALL, deciders.pick_widest(choices))
+        choices = [deciders.Depth.ATOM, deciders.Depth.FLOW,
+                   deciders.Depth.ALL, deciders.Depth.NEIGHBORS]
+        self.assertEqual(deciders.Depth.ALL, deciders.pick_widest(choices))
+        choices = [deciders.Depth.ATOM, deciders.Depth.NEIGHBORS]
+        self.assertEqual(deciders.Depth.NEIGHBORS,
+                         deciders.pick_widest(choices))
+
+    def test_bad_pick_widest(self):
+        self.assertRaises(ValueError, deciders.pick_widest, [])
+        self.assertRaises(ValueError, deciders.pick_widest, ["a"])
+        self.assertRaises(ValueError, deciders.pick_widest, set(['b']))
diff --git a/taskflow/tests/unit/test_engines.py b/taskflow/tests/unit/test_engines.py
index 288afb182..cb59a3e0e 100644
--- a/taskflow/tests/unit/test_engines.py
+++ b/taskflow/tests/unit/test_engines.py
@@ -669,6 +669,94 @@ class EngineLinearAndUnorderedExceptionsTest(utils.EngineTestBase):
         self.assertNotIn('task2.t REVERTED(None)', capturer.values)
 
 
+class EngineDeciderDepthTest(utils.EngineTestBase):
+
+    def test_run_graph_flow_decider_various_depths(self):
+        sub_flow_1 = gf.Flow('g_1')
+        g_1_1 = utils.ProgressingTask(name='g_1-1')
+        sub_flow_1.add(g_1_1)
+        g_1 = utils.ProgressingTask(name='g-1')
+        g_2 = utils.ProgressingTask(name='g-2')
+        g_3 = utils.ProgressingTask(name='g-3')
+        g_4 = utils.ProgressingTask(name='g-4')
+        for a_depth, ran_how_many in [('all', 1),
+                                      ('atom', 4),
+                                      ('flow', 2),
+                                      ('neighbors', 3)]:
+            flow = gf.Flow('g')
+            flow.add(g_1, g_2, sub_flow_1, g_3, g_4)
+            flow.link(g_1, g_2,
+                      decider=lambda history: False,
+                      decider_depth=a_depth)
+            flow.link(g_2, sub_flow_1)
+            flow.link(g_2, g_3)
+            flow.link(g_3, g_4)
+            flow.link(g_1, sub_flow_1,
+                      decider=lambda history: True,
+                      decider_depth=a_depth)
+            e = self._make_engine(flow)
+            with utils.CaptureListener(e, capture_flow=False) as capturer:
+                e.run()
+            ran_tasks = 0
+            for outcome in capturer.values:
+                if outcome.endswith("RUNNING"):
+                    ran_tasks += 1
+            self.assertEqual(ran_how_many, ran_tasks)
+
+    def test_run_graph_flow_decider_jump_over_atom(self):
+        flow = gf.Flow('g')
+        a = utils.AddOneSameProvidesRequires("a", inject={'value': 0})
+        b = utils.AddOneSameProvidesRequires("b")
+        c = utils.AddOneSameProvidesRequires("c")
+        flow.add(a, b, c, resolve_requires=False)
+        flow.link(a, b, decider=lambda history: False,
+                  decider_depth='atom')
+        flow.link(b, c)
+        e = self._make_engine(flow)
+        e.run()
+        self.assertEqual(2, e.storage.get('c'))
+        self.assertEqual(states.IGNORE, e.storage.get_atom_state('b'))
+
+    def test_run_graph_flow_decider_jump_over_bad_atom(self):
+        flow = gf.Flow('g')
+        a = utils.NoopTask("a")
+        b = utils.FailingTask("b")
+        c = utils.NoopTask("c")
+        flow.add(a, b, c)
+        flow.link(a, b, decider=lambda history: False,
+                  decider_depth='atom')
+        flow.link(b, c)
+        e = self._make_engine(flow)
+        e.run()
+
+    def test_run_graph_flow_decider_revert(self):
+        flow = gf.Flow('g')
+        a = utils.NoopTask("a")
+        b = utils.NoopTask("b")
+        c = utils.FailingTask("c")
+        flow.add(a, b, c)
+        flow.link(a, b, decider=lambda history: False,
+                  decider_depth='atom')
+        flow.link(b, c)
+        e = self._make_engine(flow)
+        with utils.CaptureListener(e, capture_flow=False) as capturer:
+            # Wrapped failure here for WBE engine, make this better in
+            # the future, perhaps via a custom testtools matcher??
+            self.assertRaises((RuntimeError, exc.WrappedFailure), e.run)
+        expected = [
+            'a.t RUNNING',
+            'a.t SUCCESS(None)',
+            'b.t IGNORE',
+            'c.t RUNNING',
+            'c.t FAILURE(Failure: RuntimeError: Woot!)',
+            'c.t REVERTING',
+            'c.t REVERTED(None)',
+            'a.t REVERTING',
+            'a.t REVERTED(None)',
+        ]
+        self.assertEqual(expected, capturer.values)
+
+
 class EngineGraphFlowTest(utils.EngineTestBase):
 
     def test_run_empty_graph_flow(self):
@@ -1043,9 +1131,10 @@ class EngineGraphConditionalFlowTest(utils.EngineTestBase):
             'task4.t IGNORE',
         ]
         self.assertEqual(expected, capturer.values)
-        self.assertEqual(1, len(histories))
-        self.assertIn('task1', histories[0])
-        self.assertIn('task2', histories[0])
+        self.assertEqual(2, len(histories))
+        for i in range(0, 2):
+            self.assertIn('task1', histories[i])
+            self.assertIn('task2', histories[i])
 
     def test_graph_flow_conditional(self):
         flow = gf.Flow('root')
@@ -1249,14 +1338,15 @@ class SerialEngineTest(EngineTaskTest,
                        EngineResetTests,
                        EngineGraphConditionalFlowTest,
                        EngineCheckingTaskTest,
+                       EngineDeciderDepthTest,
                        test.TestCase):
     def _make_engine(self, flow,
-                     flow_detail=None, store=None):
+                     flow_detail=None, store=None, **kwargs):
         return taskflow.engines.load(flow,
                                      flow_detail=flow_detail,
                                      engine='serial',
                                      backend=self.backend,
-                                     store=store)
+                                     store=store, **kwargs)
 
     def test_correct_load(self):
         engine = self._make_engine(utils.TaskNoRequiresNoReturns)
@@ -1278,11 +1368,13 @@ class ParallelEngineWithThreadsTest(EngineTaskTest,
                                     EngineMissingDepsTest,
                                     EngineGraphConditionalFlowTest,
                                     EngineCheckingTaskTest,
+                                    EngineDeciderDepthTest,
                                     test.TestCase):
     _EXECUTOR_WORKERS = 2
 
     def _make_engine(self, flow,
-                     flow_detail=None, executor=None, store=None):
+                     flow_detail=None, executor=None, store=None,
+                     **kwargs):
         if executor is None:
             executor = 'threads'
         return taskflow.engines.load(flow, flow_detail=flow_detail,
@@ -1290,7 +1382,8 @@ class ParallelEngineWithThreadsTest(EngineTaskTest,
                                      executor=executor,
                                      engine='parallel',
                                      store=store,
-                                     max_workers=self._EXECUTOR_WORKERS)
+                                     max_workers=self._EXECUTOR_WORKERS,
+                                     **kwargs)
 
     def test_correct_load(self):
         engine = self._make_engine(utils.TaskNoRequiresNoReturns)
@@ -1319,17 +1412,19 @@ class ParallelEngineWithEventletTest(EngineTaskTest,
                                      EngineMissingDepsTest,
                                      EngineGraphConditionalFlowTest,
                                      EngineCheckingTaskTest,
+                                     EngineDeciderDepthTest,
                                      test.TestCase):
 
     def _make_engine(self, flow,
-                     flow_detail=None, executor=None, store=None):
+                     flow_detail=None, executor=None, store=None,
+                     **kwargs):
         if executor is None:
             executor = futurist.GreenThreadPoolExecutor()
             self.addCleanup(executor.shutdown)
         return taskflow.engines.load(flow, flow_detail=flow_detail,
                                      backend=self.backend, engine='parallel',
                                      executor=executor,
-                                     store=store)
+                                     store=store, **kwargs)
 
 
 class ParallelEngineWithProcessTest(EngineTaskTest,
@@ -1342,6 +1437,7 @@ class ParallelEngineWithProcessTest(EngineTaskTest,
                                     EngineResetTests,
                                     EngineMissingDepsTest,
                                     EngineGraphConditionalFlowTest,
+                                    EngineDeciderDepthTest,
                                     test.TestCase):
     _EXECUTOR_WORKERS = 2
 
@@ -1350,7 +1446,8 @@ class ParallelEngineWithProcessTest(EngineTaskTest,
         self.assertIsInstance(engine, eng.ParallelActionEngine)
 
     def _make_engine(self, flow,
-                     flow_detail=None, executor=None, store=None):
+                     flow_detail=None, executor=None, store=None,
+                     **kwargs):
         if executor is None:
             executor = 'processes'
         return taskflow.engines.load(flow, flow_detail=flow_detail,
@@ -1358,7 +1455,8 @@ class ParallelEngineWithProcessTest(EngineTaskTest,
                                      engine='parallel',
                                      executor=executor,
                                      store=store,
-                                     max_workers=self._EXECUTOR_WORKERS)
+                                     max_workers=self._EXECUTOR_WORKERS,
+                                     **kwargs)
 
 
 class WorkerBasedEngineTest(EngineTaskTest,
@@ -1371,6 +1469,7 @@ class WorkerBasedEngineTest(EngineTaskTest,
                             EngineResetTests,
                             EngineMissingDepsTest,
                             EngineGraphConditionalFlowTest,
+                            EngineDeciderDepthTest,
                             test.TestCase):
     def setUp(self):
         super(WorkerBasedEngineTest, self).setUp()
@@ -1415,10 +1514,11 @@ class WorkerBasedEngineTest(EngineTaskTest,
         super(WorkerBasedEngineTest, self).tearDown()
 
     def _make_engine(self, flow,
-                     flow_detail=None, store=None):
+                     flow_detail=None, store=None, **kwargs):
+        kwargs.update(self.engine_conf)
         return taskflow.engines.load(flow, flow_detail=flow_detail,
                                      backend=self.backend,
-                                     store=store, **self.engine_conf)
+                                     store=store, **kwargs)
 
     def test_correct_load(self):
         engine = self._make_engine(utils.TaskNoRequiresNoReturns)
diff --git a/taskflow/tests/unit/test_utils.py b/taskflow/tests/unit/test_utils.py
index cf435003b..effc54b50 100644
--- a/taskflow/tests/unit/test_utils.py
+++ b/taskflow/tests/unit/test_utils.py
@@ -244,24 +244,6 @@ class TestCountdownIter(test.TestCase):
         self.assertRaises(ValueError, six.next, it)
 
 
-class TestLookFor(test.TestCase):
-    def test_no_matches(self):
-        hay = [9, 10, 11]
-        self.assertEqual([], misc.look_for(hay, [1, 2, 3]))
-
-    def test_match_order(self):
-        hay = [6, 5, 4, 3, 2, 1]
-        priors = []
-        for i in range(0, 6):
-            priors.append(i + 1)
-            matches = misc.look_for(hay, priors)
-            self.assertGreater(0, len(matches))
-            self.assertIsSuperAndSubsequence(hay, matches)
-        hay = [10, 1, 15, 3, 5, 8, 44]
-        self.assertEqual([1, 15], misc.look_for(hay, [15, 1]))
-        self.assertEqual([10, 44], misc.look_for(hay, [44, 10]))
-
-
 class TestMergeUri(test.TestCase):
     def test_merge(self):
         url = "http://www.yahoo.com/?a=b&c=d"
diff --git a/taskflow/utils/misc.py b/taskflow/utils/misc.py
index 1a516d839..5c2bf4fa5 100644
--- a/taskflow/utils/misc.py
+++ b/taskflow/utils/misc.py
@@ -226,37 +226,6 @@ def parse_uri(uri):
     return netutils.urlsplit(uri)
 
 
-def look_for(haystack, needles, extractor=None):
-    """Find items in haystack and returns matches found (in haystack order).
-
-    Given a list of items (the haystack) and a list of items to look for (the
-    needles) this will look for the needles in the haystack and returns
-    the found needles (if any). The ordering of the returned needles is in the
-    order they are located in the haystack.
-
-    Example input and output:
-
-    >>> from taskflow.utils import misc
-    >>> hay = [3, 2, 1]
-    >>> misc.look_for(hay, [1, 2])
-    [2, 1]
-    """
-    if not haystack:
-        return []
-    if extractor is None:
-        extractor = lambda v: v
-    matches = []
-    for i, v in enumerate(needles):
-        try:
-            matches.append((haystack.index(extractor(v)), i))
-        except ValueError:
-            pass
-    if not matches:
-        return []
-    else:
-        return [needles[i] for (_hay_i, i) in sorted(matches)]
-
-
 def disallow_when_frozen(excp_cls):
     """Frozen checking/raising method decorator."""