diff --git a/mani_skill2/envs/tasks/README.md b/mani_skill2/envs/tasks/README.md
index fffa92f2e..588d3c017 100644
--- a/mani_skill2/envs/tasks/README.md
+++ b/mani_skill2/envs/tasks/README.md
@@ -1,5 +1,13 @@
 # All new ManiSkill tasks
 
+To add a new task, create either a standalone python file your_task_name.py or folder (useful if you use some custom assets). We recommend using the mani_skill2/envs/template.py file to get started and filling out the code in there and following the comments.
+
+Moreover, each task is required to come along with a "Task Sheet" which quickly describes some core components of the task, namely
+- Randomization: how is the task randomized upon each environment reset?
+- Success Conditions: when is the task considered solved?
+- Visualization: Link to a gif / video of the task being solved
+
+
 Difficulty rankings (training on 4090 GPU): 
 1 - PPO takes < 2 minutes to train?
 2 - PPO takes < 5 minutes
diff --git a/mani_skill2/envs/tasks/pick_cube.py b/mani_skill2/envs/tasks/pick_cube.py
index 8df2ffb8e..76a880508 100644
--- a/mani_skill2/envs/tasks/pick_cube.py
+++ b/mani_skill2/envs/tasks/pick_cube.py
@@ -17,6 +17,8 @@
 @register_env("PickCube-v1", max_episode_steps=100)
 class PickCubeEnv(BaseEnv):
     """
+    Task Description
+    ----------------
     A simple task where the objective is to grasp a cube and move it to a target goal position.
 
     Randomizations
diff --git a/mani_skill2/envs/tasks/push_cube.py b/mani_skill2/envs/tasks/push_cube.py
index c1cbf2186..f08fbdfa9 100644
--- a/mani_skill2/envs/tasks/push_cube.py
+++ b/mani_skill2/envs/tasks/push_cube.py
@@ -37,6 +37,8 @@
 @register_env("PushCube-v0", max_episode_steps=50)
 class PushCubeEnv(BaseEnv):
     """
+    Task Description
+    ----------------
     A simple task where the objective is to push and move a cube to a goal region in front of it
 
     Randomizations
diff --git a/mani_skill2/envs/template.py b/mani_skill2/envs/template.py
index 075c59750..ba240cfca 100644
--- a/mani_skill2/envs/template.py
+++ b/mani_skill2/envs/template.py
@@ -35,6 +35,23 @@
 # register the environment by a unique ID and specify a max time limit. Now once this file is imported you can do gym.make("CustomEnv-v0")
 @register_env(name="CustomEnv-v0", max_episode_steps=200)
 class CustomEnv(BaseEnv):
+    """
+    Task Description
+    ----------------
+    Add a task description here
+
+    Randomizations
+    --------------
+    - how is it randomized?
+    - how is that randomized?
+
+    Success Conditions
+    ------------------
+    - what is done to check if this task is solved?
+
+    Visualization: link to a video/gif of the task being solved
+    """
+
     # in the __init__ function you can pick a default robot your task should use e.g. the panda robot
     def __init__(self, *args, robot_uid="panda", robot_init_qpos_noise=0.02, **kwargs):
         self.robot_init_qpos_noise = robot_init_qpos_noise