diff --git a/.gitignore b/.gitignore
index 3df8783..8f7e103 100644
--- a/.gitignore
+++ b/.gitignore
@@ -4,6 +4,6 @@
 *.pyc
 __pycache__
 
-docs/
+*.html
 *.out
 *.swp
diff --git a/docs/TODO.md b/docs/TODO.md
new file mode 100644
index 0000000..fc75cc0
--- /dev/null
+++ b/docs/TODO.md
@@ -0,0 +1,125 @@
+# TODO
+
+## Current
+
+## General
+
+* [ ] Logging
+* [ ] Testing
+* [ ] GUI
+    - [ ] Create a GUI for configuring and running the bot, and performing other jobs
+        like adding or subtracting points for specific redditors
+    - [ ] Check for updates in the Github repo and prompt user to update
+        - Github API overview:
+            - https://developer.github.com/v3/
+            - https://developer.github.com/v4/
+        - Useful links:
+            - https://developer.github.com/v3/repos/branches/#get-branch
+            - https://developer.github.com/v3/repos/commits/
+            - https://developer.github.com/v3/repos/releases/
+            - https://developer.github.com/v3/repos/hooks/
+        - Webhooks:
+            - https://developer.github.com/webhooks/
+            - https://developer.github.com/v3/activity/events/types/#pushevent
+* [ ] Determine whether and how to check "![Ss]olved" comments have been later
+    edited to remove the "![Ss]olved" string, and whether and how to remove or
+    reassign points
+    - If so, it could do that daily or something.
+    - This will be especially important if a "recovery mode" is implemented that
+        crawls through the whole subreddit to rebuild the database, since the
+        bot would only be able to see comments that haven't been removed, or
+        the newest version of edited comments.
+    - This could also just be encouraged through sub rules; e.g. "don't mark as
+        solved until you've actually tried the proposed solution"
+* [ ] As mentioned in the previous section, implement a recovery mode
+
+## File-Specific Tasks
+
+### bot.py
+
+* [ ] Allow mods and/or bot owner to add or remove points from specific users
+* [ ] Make the algorithm for determining the problem solver more sophisticated
+    - e.g. check entire comment tree instead of just ignoring if the OP also
+        authored the parent comment
+    - Ask OP for clarification only if solver cannot be sufficiently determined
+    - Again, this behavior could also perhaps be better enforced through
+        subreddit rules rather than algorithmically
+* [X] Allow mods to mark a post as solved with "/solved"
+* [X] When replying to solution, the bot should...
+    - [X] Reply to the comment containing the "![Ss]olved" string
+    - [X] Tag the solver so they will be notified
+    - ~~Delete the automod message~~
+* ~~Add a star to the user flair for every 100 points~~
+* ~~Add a way to look up user points~~
+    - i.e. summon bot by tagging it and provide a username to look up
+    - the bot will reply with the last message that the user received
+
+### config.py
+
+* [X] Change from `ini` to `toml `
+* [X] Consolidate `pointsbot.ini` and `praw.ini` into a single config file.
+* [X] Add option for flair_template_id
+* [X] Check for config file in a well-known location, e.g. `~` directory
+    - Probably named `~/.pointsbot` or something similar
+    - Could do something similar to packages like PRAW:
+        1. Look in current working directory first.
+        2. Look in OS-dependent location next.
+        3. (Maybe) Finally, could look in the directory containing the source files
+           (using `__file__`).
+* [X] Add interactive config scipt
+* ~~Fix titlecase problem with roman numerals~~
+    - Only an issue with `ini` format
+* ~~Separate the development-handy config items into separate section~~
+    - Don't really have any right now
+
+### database.py
+
+* [ ] Possibly refactor for a datastore type thing instead of database
+    - Maybe even make models like Redditor to combine data storage/access with
+        logic, e.g. determining current level
+
+### reply.py
+
+* [ ] For the footer section of the reply comment regarding the bot, could have the
+    bot make a post on its account explaining itself, and link to that (and then
+    also link to the source code separately, and perhaps in that post, too).
+    Could even have the bot make this post automatically if it doesn't have a
+    link to the post in its config, and then store the link for future use.
+* [X] Fix progress bar
+
+## Ideas
+
+### Config
+
+### Database
+
+* Should it keep track of the solved posts for future reference and calculate
+    points on the fly, rather than just keeping track of points? If so, should
+    have a column denoting whether the post has been deleted, if that
+    information is decided to be useful when determining a user's points.
+
+### Determining when to award points
+
+To ensure that points are only awarded for the first comment marked as a
+solution:
+
+* Alter the database to allow for tracking of each submission and the first
+    comment marked as the solution. Then, everytime a new solution comment is
+    detected, simply check the database to see if the submission is already
+    counted. This will avoid unnecessary calls to Reddit, which would include
+    scanning all the submission comments each time.
+* This approach could also make it simpler to check whether a solution comment
+    has been edited. Instead of having to do a daily search for edits, it could
+    just check the original solution comment to ensure that it still contains
+    the "!solved" string. If not, it can remove points from that author and
+    award points to the new author.
+
+To ensure that a point is awarded to the correct user:
+
+* We could expand the "!solved" bot summons to include an optional username
+    argument. Without the argument, the bot will award the point to either the
+    top-level comment or the parent comment of the "!solved" comment, whichever
+    is decided upon. However, if the username argument is provided, the bot
+    could simple check that one of the comments in the comment tree belongs to
+    that user, and then award them the point.
+    - Honestly, this is probably overcomplicated and unnecessary, though.
diff --git a/docs/run.sh b/docs/run.sh
new file mode 100755
index 0000000..4608071
--- /dev/null
+++ b/docs/run.sh
@@ -0,0 +1,8 @@
+#!/usr/bin/env bash
+
+docfiles=( $(ls *.md) )
+for fname in "${docfiles[@]}"; do
+    outname="${fname/%md/html}"
+    pandoc -s -f gfm -t html $fname -o $outname --metadata pagetitle="$outname"
+    open $outname
+done