diff --git a/python/completers/all/identifier_completer.py b/python/completers/all/identifier_completer.py
index fcd3f07f..ba0e82e0 100644
--- a/python/completers/all/identifier_completer.py
+++ b/python/completers/all/identifier_completer.py
@@ -17,32 +17,24 @@
 # You should have received a copy of the GNU General Public License
 # along with YouCompleteMe.  If not, see <http://www.gnu.org/licenses/>.
 
-from completers.completer import Completer
+from completers.completer import GeneralCompleter
 import vim
 import vimsupport
 import ycm_core
 import ycm_utils as utils
 
 MAX_IDENTIFIER_COMPLETIONS_RETURNED = 10
-MIN_NUM_CHARS = int( vimsupport.GetVariableValue(
-  "g:ycm_min_num_of_chars_for_completion" ) )
 
 
-class IdentifierCompleter( Completer ):
+class IdentifierCompleter( GeneralCompleter ):
   def __init__( self ):
     super( IdentifierCompleter, self ).__init__()
     self.completer = ycm_core.IdentifierCompleter()
     self.completer.EnableThreading()
 
 
-  def SupportedFiletypes( self ):
-    # magic token meaning all filetypes
-    return set( [ 'ycm_all' ] )
-
-
   def ShouldUseNow( self, start_column ):
-    query_length = vimsupport.CurrentColumn() - start_column
-    return query_length >= MIN_NUM_CHARS
+      return self.QueryLengthAboveMinThreshold( start_column )
 
 
   def CandidatesForQueryAsync( self, query, unused_start_column ):
diff --git a/python/completers/completer.py b/python/completers/completer.py
index e5e18675..9a3868de 100644
--- a/python/completers/completer.py
+++ b/python/completers/completer.py
@@ -21,11 +21,14 @@ import abc
 import vim
 import vimsupport
 import ycm_core
-import ycm
 from collections import defaultdict
+from threading import Event
 
 NO_USER_COMMANDS = 'This completer does not define any commands.'
 
+MIN_NUM_CHARS = int( vimsupport.GetVariableValue(
+  "g:ycm_min_num_of_chars_for_completion" ) )
+
 class Completer( object ):
   """A base class for all Completers in YCM.
 
@@ -154,6 +157,11 @@ class Completer( object ):
     return False
 
 
+  def QueryLengthAboveMinThreshold( self, start_column ):
+    query_length = vimsupport.CurrentColumn() - start_column
+    return query_length >= MIN_NUM_CHARS
+
+
   # It's highly likely you DON'T want to override this function but the *Inner
   # version of it.
   def CandidatesForQueryAsync( self, query, start_column ):
@@ -178,10 +186,15 @@ class Completer( object ):
       candidates = candidates.words
     items_are_objects = 'word' in candidates[ 0 ]
 
-    return ycm_core.FilterAndSortCandidates(
-      candidates,
-      'word' if items_are_objects else '',
-      query )
+    try:
+      matches = ycm_core.FilterAndSortCandidates(
+        candidates,
+        'word' if items_are_objects else '',
+        query )
+    except:
+      matches = []
+
+    return matches
 
 
   def CandidatesForQueryAsyncInner( self, query, start_column ):
@@ -295,6 +308,31 @@ class Completer( object ):
     return ''
 
 
+class GeneralCompleter( Completer ):
+  """
+  A base class for General completers in YCM.
+
+  Because this is a subclass of Completer class, you should refer to the
+  dpcumentation of Completer API.
+
+  Only exception is that GeneralCompleterStore class that collects and controls
+  all general completers already adds threading for completers, so there
+  is no need to add a threading to new general completers.
+
+  added __init__ fields are for GeneralCompleterStore internal use only.
+  """
+  def __init__( self ):
+    super( GeneralCompleter, self ).__init__()
+    self._should_start = Event()
+    self._should_use = False
+    self._finished = Event()
+    self._results = []
+
+
+  def SupportedFiletypes( self ):
+    return set()
+
+
 class CompletionsCache( object ):
   def __init__( self ):
     self.line = -1
diff --git a/python/completers/general/filename_completer.py b/python/completers/general/filename_completer.py
new file mode 100644
index 00000000..f1d1d158
--- /dev/null
+++ b/python/completers/general/filename_completer.py
@@ -0,0 +1,94 @@
+#!/usr/bin/env python
+#
+# Copyright (C) 2013 Stanislav Golovanov <stgolovanov@gmail.com>
+#
+#
+# YouCompleteMe is free software: you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation, either version 3 of the License, or
+# (at your option) any later version.
+#
+# YouCompleteMe is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with YouCompleteMe.  If not, see <http://www.gnu.org/licenses/>.
+
+from completers.completer import GeneralCompleter, CompletionsCache
+import vimsupport
+import ycm_core
+import vim
+import os
+import re
+
+
+class FilenameCompleter( GeneralCompleter ):
+  """
+  General completer that provides filename and filepath completions.
+
+  It maintains a cache of completions which is invalidated on each '/' symbol.
+  """
+  def __init__(self):
+    super( FilenameCompleter, self ).__init__()
+    self._candidates = []
+    self._query = None
+    self._should_use = False
+
+    # TODO look into vim-style path globbing, NCC has a nice implementation
+    self._path_regex = re.compile( """(?:[A-z]+:/|[/~]|\./|\.+/)+ # 1 or more 'D:/'-like token or '/' or '~' or './' or '../'
+                             (?:[ /a-zA-Z0-9()$+_~.\x80-\xff-\[\]]| # any alphanumeric symbal and space literal
+                             [^\x20-\x7E]| # skip any special symbols
+                             \\.)* # backslash and 1 char after it. + matches 1 or more of whole group
+                           """, re.X )
+
+
+  def ShouldUseNowInner( self, start_column ):
+    token = vim.current.line[ start_column - 1 ]
+    if token  == '/' or self._should_use:
+      self._should_use = True
+      return True
+    else:
+      return False
+
+
+  def CandidatesForQueryAsyncInner( self, query, start_column ):
+    self._candidates = []
+    self._query = query
+    self._completions_ready = False
+    self.line = str( vim.current.line.strip() )
+    self.SetCandidates()
+
+
+  def AsyncCandidateRequestReadyInner( self ):
+    return self._completions_ready
+
+
+  def OnInsertLeave( self ):
+    # TODO this a hackish way to keep results when typing 2-3rd char after slash
+    # because identifier completer will kick in and replace results for 1 char.
+    # Need to do something better
+    self._should_use = False
+
+
+  def CandidatesFromStoredRequestInner( self ):
+    return self._candidates
+
+
+  def SetCandidates( self ):
+    path = self._path_regex.search( self.line )
+    self._working_dir = os.path.expanduser( path.group() ) if path else ''
+
+    try:
+      paths = os.listdir( self._working_dir )
+    except:
+      paths = []
+
+    self._candidates = [ {'word': path,
+                        'dup': 1,
+                        'menu': '[Dir]' if os.path.isdir( self._working_dir + \
+                                                         '/' + path ) else '[File]'
+                        } for path in paths ]
+
+    self._completions_ready = True
diff --git a/python/completers/general/general_completer.py b/python/completers/general/general_completer.py
index 25a0c243..d625c8fe 100644
--- a/python/completers/general/general_completer.py
+++ b/python/completers/general/general_completer.py
@@ -19,7 +19,7 @@
 
 from completers.completer import Completer
 from completers.all.identifier_completer import IdentifierCompleter
-from threading import Thread, Event
+from threading import Thread
 import vimsupport
 import inspect
 import fnmatch
@@ -58,11 +58,10 @@ class GeneralCompleterStore( Completer ):
   def InitCompleters( self ):
     # This method creates objects of main completers class.
     completers = []
-    modules = [ module for module in os.listdir( '.' )
+    modules = [ module for module in os.listdir( os.path.dirname(__file__) )
                 if fnmatch.fnmatch(module, '*.py')
                 and not 'general_completer' in module
-                and not '__init__' in module
-                and not 'hook' in module ]
+                and not '__init__' in module ]
 
     for module in modules:
 
@@ -79,20 +78,21 @@ class GeneralCompleterStore( Completer ):
 
       for _, ClassObject in inspect.getmembers( module, inspect.isclass ):
         # Iterate over all classes in a module and select main class
-        if hasattr( ClassObject, 'CandidatesForQueryAsyncInner' ):
+        if not __name__ in str(ClassObject) and 'general' in str(ClassObject):
+
           classInstance = ClassObject
 
-      # Init selected class and store class object
-      completers.append( GeneralCompleterInstance( classInstance() ) )
 
-    # append IdentifierCompleter
-    completers.append( GeneralCompleterInstance( IdentifierCompleter() ) )
+      # Init selected class and store class object
+      completers.append( classInstance() )
+    
+    completers.append( IdentifierCompleter() )
+
     return completers
 
 
   def SupportedFiletypes( self ):
-    # magic token meaning all filetypes
-    return set( [ 'ycm_all' ] )
+    return set()
 
 
   def ShouldUseNow( self, start_column ):
@@ -100,9 +100,9 @@ class GeneralCompleterStore( Completer ):
     # True. Also update flags in completers classes
     flag = False
     for completer in self.completers:
-      ShouldUse = completer.completer.ShouldUseNow( start_column )
-      completer.ShouldUse = ShouldUse
-      if ShouldUse:
+      _should_use = completer.ShouldUseNow( start_column )
+      completer._should_use = _should_use
+      if _should_use:
         flag = True
 
     return flag
@@ -114,23 +114,23 @@ class GeneralCompleterStore( Completer ):
 
     # if completer should be used start thread by setting Event flag
     for completer in self.completers:
-      completer.finished.clear()
-      if completer.ShouldUse and not completer.should_start.is_set():
-        completer.should_start.set()
+      completer._finished.clear()
+      if completer._should_use and not completer._should_start.is_set():
+        completer._should_start.set()
 
 
   def AsyncCandidateRequestReady( self ):
     # Return True when all completers that should be used are finished their work.
     for completer in self.completers:
-        if not completer.finished.is_set() and completer.ShouldUse:
+        if not completer._finished.is_set() and completer._should_use:
             return False
     return True
 
 
   def CandidatesFromStoredRequest( self ):
     for completer in self.completers:
-      if completer.ShouldUse and completer.finished.is_set():
-          self._candidates += completer.results.pop()
+      if completer._should_use and completer._finished.is_set():
+          self._candidates += completer._results.pop()
 
     return self._candidates
 
@@ -139,17 +139,17 @@ class GeneralCompleterStore( Completer ):
     while True:
 
       # sleep until ShouldUseNow returns True
-      WaitAndClear( completer.should_start )
+      WaitAndClear( completer._should_start )
 
-      completer.completer.CandidatesForQueryAsync( self.query,
+      completer.CandidatesForQueryAsync( self.query,
                                                    self.completion_start_column )
 
-      while not completer.completer.AsyncCandidateRequestReady():
+      while not completer.AsyncCandidateRequestReady():
           continue
 
-      completer.results.append( completer.completer.CandidatesFromStoredRequest() )
+      completer._results.append( completer.CandidatesFromStoredRequest() )
 
-      completer.finished.set()
+      completer._finished.set()
 
 
   def StartThreads( self ):
@@ -161,50 +161,38 @@ class GeneralCompleterStore( Completer ):
     # Process all parsing methods of completers. Needed by identifier completer
     for completer in self.completers:
       # clear all stored completion results
-      completer.results = []
-      completer.completer.OnFileReadyToParse()
+      completer._results = []
+      completer.OnFileReadyToParse()
 
 
   def OnCursorMovedInsertMode( self ):
     for completer in self.completers:
-      completer.completer.OnCursorMovedInsertMode()
+      completer.OnCursorMovedInsertMode()
 
 
   def OnCursorMovedNormalMode( self ):
     for completer in self.completers:
-      completer.completer.OnCursorMovedNormalMode()
+      completer.OnCursorMovedNormalMode()
 
 
   def OnBufferVisit( self ):
     for completer in self.completers:
-      completer.completer.OnBufferVisit()
+      completer.OnBufferVisit()
 
 
   def OnBufferDelete( self, deleted_buffer_file ):
     for completer in self.completers:
-      completer.completer.OnBufferDelete( deleted_buffer_file )
+      completer.OnBufferDelete( deleted_buffer_file )
 
 
   def OnCursorHold( self ):
     for completer in self.completers:
-      completer.completer.OnCursorHold()
+      completer.OnCursorHold()
 
 
   def OnInsertLeave( self ):
     for completer in self.completers:
-      completer.completer.OnInsertLeave()
-
-
-class GeneralCompleterInstance( object ):
-  """
-  Class that holds all meta information about specific general completer
-  """
-  def __init__( self, completer ):
-    self.completer = completer
-    self.should_start = Event()
-    self.ShouldUse = False
-    self.finished = Event()
-    self.results = []
+      completer.OnInsertLeave()
 
 
 def WaitAndClear( event, timeout=None ):
diff --git a/python/completers/general/hook.py b/python/completers/general/hook.py
deleted file mode 100644
index e84cf950..00000000
--- a/python/completers/general/hook.py
+++ /dev/null
@@ -1,21 +0,0 @@
-# Copyright (C) 2013 Stanislav Golovanov <stgolovanov@gmail.com>
-#
-# This file is part of YouCompleteMe.
-#
-# YouCompleteMe is free software: you can redistribute it and/or modify
-# it under the terms of the GNU General Public License as published by
-# the Free Software Foundation, either version 3 of the License, or
-# (at your option) any later version.
-#
-# YouCompleteMe is distributed in the hope that it will be useful,
-# but WITHOUT ANY WARRANTY; without even the implied warranty of
-# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-# GNU General Public License for more details.
-#
-# You should have received a copy of the GNU General Public License
-# along with YouCompleteMe.  If not, see <http://www.gnu.org/licenses/>.
-
-from general_completer import GeneralCompleterStore
-
-def GetCompleter():
-    return GeneralCompleterStore()
diff --git a/python/completers/general/ultisnips_completer.py b/python/completers/general/ultisnips_completer.py
index 367e4b08..1c31f7de 100644
--- a/python/completers/general/ultisnips_completer.py
+++ b/python/completers/general/ultisnips_completer.py
@@ -17,14 +17,12 @@
 # You should have received a copy of the GNU General Public License
 # along with YouCompleteMe.  If not, see <http://www.gnu.org/licenses/>.
 
-from completers.completer import Completer, CompletionsCache
+from completers.completer import GeneralCompleter, CompletionsCache
 from UltiSnips import UltiSnips_Manager
 import vimsupport
 
-MIN_NUM_CHARS = int( vimsupport.GetVariableValue(
-  "g:ycm_min_num_of_chars_for_completion" ) )
 
-class UltiSnipsCompleter( Completer ):
+class UltiSnipsCompleter( GeneralCompleter ):
   """
   General completer that provides UltiSnips snippet names in completions.
 
@@ -47,15 +45,11 @@ class UltiSnipsCompleter( Completer ):
 
 
   def ShouldUseNowInner( self, start_column ):
-    query_length = vimsupport.CurrentColumn() - start_column
-    return query_length >= MIN_NUM_CHARS
-
-
-  def SupportedFiletypes( self ):
-    # magic token meaning all filetypes
-    return set( [ 'ycm_all' ] )
+    return self.QueryLengthAboveMinThreshold( start_column )
 
 
+  # We need to override this because Completer version invalidates cache on
+  # empty query and we want to invalidate cache only on buffer switch.
   def CandidatesForQueryAsync( self, query, start_column ):
     self.completion_start_column = start_column
 
@@ -77,27 +71,17 @@ class UltiSnipsCompleter( Completer ):
     return self.flag
 
 
-  # We need to override this because we need to store all snippets but return
-  # filtered results on first call.
-  def CandidatesFromStoredRequest( self ):
-    if self.completions_cache:
-      return self.completions_cache.filtered_completions
-    else:
-      self.completions_cache = CompletionsCache()
-      self.completions_cache.raw_completions = self.CandidatesFromStoredRequestInner()
-      self.completions_cache.line, _ = vimsupport.CurrentLineAndColumn()
-      self.completions_cache.column = self.completion_start_column
-      return self.FilterAndSortCandidates( self._candidates, self._query )
-
-
   def CandidatesFromStoredRequestInner( self ):
     return self._candidates
 
 
   def SetCandidates( self ):
     try:
-      # get all snippets for filetype
       rawsnips = UltiSnips_Manager._snips( '', 1 )
+
+      # UltiSnips_Manager._snips() returns a class instance where:
+      # class.trigger - name of snippet trigger word ( e.g. defn or testcase )
+      # class.description - description of the snippet
       self._candidates = [ { 'word': str( snip.trigger ),
                               'menu': str( '<snip> ' + snip.description ) }
                           for snip in rawsnips ]
@@ -108,5 +92,6 @@ class UltiSnipsCompleter( Completer ):
 
   def OnFileReadyToParse( self ):
     # Invalidate cache on buffer switch
-    self.completions_cache = None
+    self.completions_cache = CompletionsCache()
     self.SetCandidates()
+    self.completions_cache.raw_completions = self._candidates