diff --git a/docs/README.md b/docs/README.md
index 9af49c91..d0be1d5c 100644
--- a/docs/README.md
+++ b/docs/README.md
@@ -17,6 +17,7 @@ Note: If you'd like to customize any of the spoken forms, please see the [docume
     - [Marks](#marks)
       - [Decorated symbol](#decorated-symbol)
         - [Colors](#colors)
+        - [Shapes](#shapes)
       - [`"this"`](#this)
       - [`"that"`](#that)
     - [Modifiers](#modifiers)
@@ -82,32 +83,50 @@ Combining this with an action, we might say `"take blue air"` to select the toke
 
 The following colors are supported:
 
-| Command    | Visible color |
-| ---------- | ------------- |
-| `"blue"`   | Blue          |
-| `"green"`  | Green         |
-| `"rose"`   | Red           |
-| `"squash"` | Yellow        |
-| `"plum"`   | Pink          |
+| Spoken form | Visible color | Internal ID |
+| ----------- | ------------- | ----------- |
+| `"blue"`    | Blue          | `blue`      |
+| `"green"`   | Green         | `green`     |
+| `"rose"`    | Red           | `rose`      |
+| `"squash"`  | Yellow        | `yellow`    |
+| `"plum"`    | Pink          | `pink`      |
 
-###### Shapes
+You can enable or disable colors in your VSCode settings, by searching for `cursorless.hatEnablement.colors` and checking the box next to the internal ID for the given shape as listed above.
 
-The following shapes are supported:
+You can also tweak the visible colors for any of these colors in your VSCode settings, by searching for `cursorless.colors` and changing the hex color code next to the internal ID for the given shape as listed above. Note that you can configure different colors for dark and light themes.
 
-| Command   | Visible color    | Enabled by default? |
-| --------- | ---------------- | ------------------- |
-| `"splat"` | Four-point star  | ✅                  |
-| `"fox"`   | Chevron          | ✅                  |
-| `"wing"`  | Three-point star | ❌                  |
-| `"hole"`  | Hole             | ❌                  |
-| `"frame"` | Frame            | ❌                  |
-| `"curve"` | Curve            | ❌                  |
-| `"stare"` | Eye              | ❌                  |
+If you find these color names unintuitive / tough to remember, their
+spoken forms can be [customized](customization.md) like any other spoken form
+in cursorless. If you change a spoken form to be more than one syllable, you
+can change the penalty in the `cursorless.hatPenalties.colors` setting to the
+number of syllables you use, so that cursorless can optimize hat allocation to
+minimize syllables.
 
-To enable or disable shapes requires the following two steps:
+###### Shapes
+
+The following shapes are supported:
 
-1. Check the box corresponding to the given shape in the `cursorless.hatEnablement.shapes` field of the cursorless vscode settings
-2. Enable the corresponding spoken form in the [spoken form customization csvs](customization.md) for cursorless talon
+| Spoken form | Internal ID | Shape                                                                                          | Enabled by default? |
+| ----------- | ----------- | ---------------------------------------------------------------------------------------------- | ------------------- |
+| `"ex"`      | `ex`        | ![Ex](https://raw.githubusercontent.com/pokey/cursorless-vscode/main/images/hats/ex.svg)       | ❌                  |
+| `"fox"`     | `fox`       | ![Fox](https://raw.githubusercontent.com/pokey/cursorless-vscode/main/images/hats/fox.svg)     | ❌                  |
+| `"wing"`    | `wing`      | ![Wing](https://raw.githubusercontent.com/pokey/cursorless-vscode/main/images/hats/wing.svg)   | ❌                  |
+| `"hole"`    | `hole`      | ![Hole](https://raw.githubusercontent.com/pokey/cursorless-vscode/main/images/hats/hole.svg)   | ❌                  |
+| `"frame"`   | `frame`     | ![Frame](https://raw.githubusercontent.com/pokey/cursorless-vscode/main/images/hats/frame.svg) | ❌                  |
+| `"curve"`   | `curve`     | ![Curve](https://raw.githubusercontent.com/pokey/cursorless-vscode/main/images/hats/curve.svg) | ❌                  |
+| `"eye"`     | `eye`       | ![Eye](https://raw.githubusercontent.com/pokey/cursorless-vscode/main/images/hats/eye.svg)     | ❌                  |
+| `"play"`    | `play`      | ![Play](https://raw.githubusercontent.com/pokey/cursorless-vscode/main/images/hats/play.svg)   | ❌                  |
+| `"star"`    | `star`      | ![Star](https://raw.githubusercontent.com/pokey/cursorless-vscode/main/images/hats/star.svg)   | ❌                  |
+| `"bolt"`    | `bolt`      | ![Bolt](https://raw.githubusercontent.com/pokey/cursorless-vscode/main/images/hats/bolt.svg)   | ❌                  |
+
+You can enable or disable shapes in your VSCode settings, by searching for `cursorless.hatEnablement.shapes` and checking the box next to the internal ID for the given shape as listed above.
+
+If you find these shape names unintuitive / tough to remember, their
+spoken forms can be [customized](customization.md) like any other spoken form
+in cursorless. If you change a spoken form to be more than one syllable, you
+can change the penalty in the `cursorless.hatPenalties.shapes` setting to the
+number of syllables you use, so that cursorless can optimize hat allocation to
+minimize syllables.
 
 ##### `"this"`
 
diff --git a/src/cheat_sheet.py b/src/cheat_sheet.py
index 87913d27..f6ca6d32 100644
--- a/src/cheat_sheet.py
+++ b/src/cheat_sheet.py
@@ -203,13 +203,17 @@ def draw(self, canvas):
             },
         )
 
-        self.next_row()
-        self.draw_header(canvas, "Colors")
-        self.draw_items(canvas, get_list("hat_color"))
-
-        self.next_row()
-        self.draw_header(canvas, "Shapes")
-        self.draw_items(canvas, get_list("hat_shape"))
+        hat_colors = get_list("hat_color")
+        if hat_colors:
+            self.next_row()
+            self.draw_header(canvas, "Colors")
+            self.draw_items(canvas, hat_colors)
+
+        hat_shapes = get_list("hat_shape")
+        if hat_shapes:
+            self.next_row()
+            self.draw_header(canvas, "Shapes")
+            self.draw_items(canvas, hat_shapes)
 
         self.next_row()
         self.draw_header(canvas, "Examples")
diff --git a/src/csv_overrides.py b/src/csv_overrides.py
index 88744057..ed004b45 100644
--- a/src/csv_overrides.py
+++ b/src/csv_overrides.py
@@ -14,7 +14,11 @@
 )
 
 
-def init_csv_and_watch_changes(filename: str, default_values: dict[str, dict]):
+def init_csv_and_watch_changes(
+    filename: str,
+    default_values: dict[str, dict],
+    extra_acceptable_values: list[str] = None,
+):
     """
     Initialize a cursorless settings csv, creating it if necessary, and watch
     for changes to the csv.  Talon lists will be generated based on the keys of
@@ -33,7 +37,12 @@ def init_csv_and_watch_changes(filename: str, default_values: dict[str, dict]):
         `cursorles-settings` dir
         default_values (dict[str, dict]): The default values for the lists to
         be customized in the given csv
+        extra_acceptable_values list[str]: Don't throw an exception if any of
+        these appear as values
     """
+    if extra_acceptable_values is None:
+        extra_acceptable_values = []
+
     dir_path, file_path = get_file_paths(filename)
     super_default_values = get_super_values(default_values)
 
@@ -42,25 +51,36 @@ def init_csv_and_watch_changes(filename: str, default_values: dict[str, dict]):
     def on_watch(path, flags):
         if file_path.match(path):
             current_values, has_errors = read_file(
-                file_path, super_default_values.values()
+                file_path, super_default_values.values(), extra_acceptable_values
             )
-            update_dicts(default_values, current_values)
+            update_dicts(default_values, current_values, extra_acceptable_values)
 
     fs.watch(dir_path, on_watch)
 
     if file_path.is_file():
-        current_values = update_file(file_path, super_default_values)
-        update_dicts(default_values, current_values)
+        current_values = update_file(
+            file_path, super_default_values, extra_acceptable_values
+        )
+        update_dicts(default_values, current_values, extra_acceptable_values)
     else:
         create_file(file_path, super_default_values)
-        update_dicts(default_values, super_default_values)
+        update_dicts(default_values, super_default_values, extra_acceptable_values)
+
+    def unsubscribe():
+        fs.unwatch(dir_path, on_watch)
+
+    return unsubscribe
 
 
 def is_removed(value: str):
     return value.startswith("-")
 
 
-def update_dicts(default_values: dict[str, dict], current_values: dict):
+def update_dicts(
+    default_values: dict[str, dict],
+    current_values: dict,
+    extra_acceptable_values: list[str],
+):
     # Create map with all default values
     results_map = {}
     for list_name, dict in default_values.items():
@@ -69,7 +89,13 @@ def update_dicts(default_values: dict[str, dict], current_values: dict):
 
     # Update result with current values
     for key, value in current_values.items():
-        results_map[value]["key"] = key
+        try:
+            results_map[value]["key"] = key
+        except KeyError:
+            if value in extra_acceptable_values:
+                pass
+            else:
+                raise
 
     # Convert result map back to result list
     results = {key: {} for key in default_values}
@@ -84,8 +110,10 @@ def update_dicts(default_values: dict[str, dict], current_values: dict):
         ctx.lists[get_cursorless_list_name(list_name)] = dict
 
 
-def update_file(path: Path, default_values: dict):
-    current_values, has_errors = read_file(path, default_values.values())
+def update_file(path: Path, default_values: dict, extra_acceptable_values: list[str]):
+    current_values, has_errors = read_file(
+        path, default_values.values(), extra_acceptable_values
+    )
     current_identifiers = current_values.values()
 
     missing = {}
@@ -149,7 +177,9 @@ def csv_error(path: Path, index: int, message: str, value: str):
     print(f"ERROR: {path}:{index+1}: {message} '{value}'")
 
 
-def read_file(path: Path, default_identifiers: list[str]):
+def read_file(
+    path: Path, default_identifiers: list[str], extra_acceptable_values: list[str]
+):
     with open(path) as f:
         lines = list(f)
 
@@ -180,7 +210,7 @@ def read_file(path: Path, default_identifiers: list[str]):
             seen_header = True
             continue
 
-        if value not in default_identifiers:
+        if value not in default_identifiers and value not in extra_acceptable_values:
             has_errors = True
             csv_error(path, i, "Unknown identifier", value)
             continue
diff --git a/src/marks/mark.py b/src/marks/mark.py
index 8e4fba78..ddba246b 100644
--- a/src/marks/mark.py
+++ b/src/marks/mark.py
@@ -1,6 +1,7 @@
 from dataclasses import dataclass
+from pathlib import Path
 from ..conventions import get_cursorless_list_name
-from talon import Module, app, Context
+from talon import Module, actions, app, Context, fs, cron
 from ..csv_overrides import init_csv_and_watch_changes
 
 mod = Module()
@@ -21,13 +22,16 @@
 }
 
 hat_shapes = {
-    "splat": "fourPointStar",
-    "fox": "chevron",
-    "-wing": "threePointStar",
-    "-hole": "hole",
-    "-frame": "frame",
-    "-curve": "curve",
-    "-stare": "eye",
+    "ex": "ex",
+    "fox": "fox",
+    "wing": "wing",
+    "hole": "hole",
+    "frame": "frame",
+    "curve": "curve",
+    "eye": "eye",
+    "play": "play",
+    "star": "star",
+    "bolt": "bolt",
 }
 
 
@@ -97,20 +101,61 @@ def cursorless_mark(m) -> str:
     return m.cursorless_line_number_simple
 
 
-def on_ready():
-    init_csv_and_watch_changes(
-        "special_marks",
+unsubscribe_hat_styles = None
+
+
+color_enablements = set()
+shape_enablements = set()
+
+
+def setup_hat_styles_csv(enablements: dict):
+    global unsubscribe_hat_styles, color_enablements, shape_enablements
+
+    new_color_enablements = set(enablements["colors"])
+    new_shape_enablements = set(enablements["shapes"])
+
+    if (
+        color_enablements == new_color_enablements
+        and shape_enablements == new_shape_enablements
+    ):
+        return
+
+    shape_enablements = new_shape_enablements
+    color_enablements = new_color_enablements
+
+    active_hat_colors = {
+        spoken_form: value
+        for spoken_form, value in hat_colors.items()
+        if value in enablements["colors"]
+    }
+    active_hat_shapes = {
+        spoken_form: value
+        for spoken_form, value in hat_shapes.items()
+        if value in enablements["shapes"]
+    }
+
+    if unsubscribe_hat_styles is not None:
+        unsubscribe_hat_styles()
+
+    unsubscribe_hat_styles = init_csv_and_watch_changes(
+        "hat_styles",
         {
-            "special_mark": special_marks_defaults,
+            "hat_color": active_hat_colors,
+            "hat_shape": active_hat_shapes,
         },
+        [*hat_colors.values(), *hat_shapes.values()],
     )
+
+
+def on_ready():
     init_csv_and_watch_changes(
-        "hat_styles",
+        "special_marks",
         {
-            "hat_color": hat_colors,
-            "hat_shape": hat_shapes,
+            "special_mark": special_marks_defaults,
         },
     )
 
+    actions.user.watch_vscode_state("cursorless.hatEnablement", setup_hat_styles_csv)
+
 
 app.register("ready", on_ready)
diff --git a/src/marks/vscode_settings.py b/src/marks/vscode_settings.py
new file mode 100644
index 00000000..43b0954a
--- /dev/null
+++ b/src/marks/vscode_settings.py
@@ -0,0 +1,58 @@
+import json
+import os
+from talon import Context, Module, actions, cron
+from pathlib import Path
+from ..vendor.jstyleson import loads
+
+mod = Module()
+
+windows_ctx = Context()
+mac_ctx = Context()
+linux_ctx = Context()
+
+windows_ctx.matches = r"""
+os: windows
+"""
+mac_ctx.matches = r"""
+os: mac
+"""
+linux_ctx.matches = r"""
+os: linux
+"""
+
+
+@mod.action_class
+class Actions:
+    def vscode_settings_path() -> Path:
+        """Get path of vscode settings json file"""
+        pass
+
+    def vscode_get_setting(key: str, default_value: any = None):
+        """Get the value of vscode setting at the given key"""
+        path: Path = actions.user.vscode_settings_path()
+        settings: dict = loads(path.read_text())
+
+        if default_value is not None:
+            return settings.get(key, default_value)
+        else:
+            return settings[key]
+
+
+@mac_ctx.action_class("user")
+class MacUserActions:
+    def vscode_settings_path() -> Path:
+        return Path(
+            f"{os.environ['HOME']}/Library/Application Support/Code/User/settings.json"
+        )
+
+
+@linux_ctx.action_class("user")
+class LinuxUserActions:
+    def vscode_settings_path() -> Path:
+        return Path(f"{os.environ['HOME']}/.config/Code/User/settings.json")
+
+
+@windows_ctx.action_class("user")
+class WindowsUserActions:
+    def vscode_settings_path() -> Path:
+        return Path(f"{os.environ['APPDATA']}/Code/User/settings.json")
diff --git a/src/vendor/__init__.py b/src/vendor/__init__.py
new file mode 100644
index 00000000..e69de29b
diff --git a/src/vendor/jstyleson.py b/src/vendor/jstyleson.py
new file mode 100644
index 00000000..fe1bbf1c
--- /dev/null
+++ b/src/vendor/jstyleson.py
@@ -0,0 +1,137 @@
+# From https://github.com/linjackson78/jstyleson/blob/8c47cc9e665b3b1744cccfaa7a650de5f3c575dd/jstyleson.py
+# License https://github.com/linjackson78/jstyleson/blob/8c47cc9e665b3b1744cccfaa7a650de5f3c575dd/LICENSE
+import json
+
+
+def dispose(json_str):
+    """Clear all comments in json_str.
+
+    Clear JS-style comments like // and /**/ in json_str.
+    Accept a str or unicode as input.
+
+    Args:
+        json_str: A json string of str or unicode to clean up comment
+
+    Returns:
+        str: The str without comments (or unicode if you pass in unicode)
+    """
+    result_str = list(json_str)
+    escaped = False
+    normal = True
+    sl_comment = False
+    ml_comment = False
+    quoted = False
+
+    a_step_from_comment = False
+    a_step_from_comment_away = False
+
+    former_index = None
+
+    for index, char in enumerate(json_str):
+
+        if escaped:  # We have just met a '\'
+            escaped = False
+            continue
+
+        if a_step_from_comment:  # We have just met a '/'
+            if char != '/' and char != '*':
+                a_step_from_comment = False
+                normal = True
+                continue
+
+        if a_step_from_comment_away:  # We have just met a '*'
+            if char != '/':
+                a_step_from_comment_away = False
+
+        if char == '"':
+            if normal and not escaped:
+                # We are now in a string
+                quoted = True
+                normal = False
+            elif quoted and not escaped:
+                # We are now out of a string
+                quoted = False
+                normal = True
+
+        elif char == '\\':
+            # '\' should not take effect in comment
+            if normal or quoted:
+                escaped = True
+
+        elif char == '/':
+            if a_step_from_comment:
+                # Now we are in single line comment
+                a_step_from_comment = False
+                sl_comment = True
+                normal = False
+                former_index = index - 1
+            elif a_step_from_comment_away:
+                # Now we are out of comment
+                a_step_from_comment_away = False
+                normal = True
+                ml_comment = False
+                for i in range(former_index, index + 1):
+                    result_str[i] = ""
+
+            elif normal:
+                # Now we are just one step away from comment
+                a_step_from_comment = True
+                normal = False
+
+        elif char == '*':
+            if a_step_from_comment:
+                # We are now in multi-line comment
+                a_step_from_comment = False
+                ml_comment = True
+                normal = False
+                former_index = index - 1
+            elif ml_comment:
+                a_step_from_comment_away = True
+        elif char == '\n':
+            if sl_comment:
+                sl_comment = False
+                normal = True
+                for i in range(former_index, index + 1):
+                    result_str[i] = ""
+        elif char == ']' or char == '}':
+            if normal:
+                _remove_last_comma(result_str, index)
+
+    #  To remove single line comment which is the last line of json
+    if sl_comment:
+        sl_comment = False
+        normal = True
+        for i in range(former_index, len(json_str)):
+            result_str[i] = ""
+
+    # Show respect to original input if we are in python2
+    return ("" if isinstance(json_str, str) else u"").join(result_str)
+
+
+# There may be performance suffer backtracking the last comma
+def _remove_last_comma(str_list, before_index):
+    i = before_index - 1
+    while str_list[i].isspace() or not str_list[i]:
+        i -= 1
+
+    # This is the first none space char before before_index
+    if str_list[i] == ',':
+        str_list[i] = ''
+
+
+# Below are just some wrapper function around the standard json module.
+
+def loads(text, **kwargs):
+    return json.loads(dispose(text), **kwargs)
+
+
+def load(fp, **kwargs):
+    return loads(fp.read(), **kwargs)
+
+
+def dumps(obj, **kwargs):
+    return json.dumps(obj, **kwargs)
+
+
+def dump(obj, fp, **kwargs):
+    json.dump(obj, fp, **kwargs)
\ No newline at end of file