These are some of the future works I had planned, but couldn't complete during GSoC 2017.

• Wumpus World: HybridWumpusAgent needs to be implemented. After this we can display the environment, Wumpus, agent and percepts in the notebook for a better visualization.
• Eliminate Dependencies: The base code is mostly independent of external libraries but uses version specific python functions. Such functions (absent or behave differently in older version) need to be identified and fixed. Dependencies like 'networkx' are used in the notebooks and can be replaced by adding code to notebook.py which replicates the required functionality. TravisCI testing can also be sped up by excluding unnecessary modules like IPython from travis setup.
• Expression Parsing: Currently Expr supports having variable number of arguments as a list but functions like unify() do not consider nor expr make use of this fact.
• Deploy Notebooks: Run the jupyter notebooks on platforms like binder or docker so that it is accessible online to everyone.
• Notebook Examples: Many algorithms are yet to be added to the notebooks. The notebooks should include a description of what it does and a few examples of the algorithm being used. Fitting visualization also help a lot.