LeetBot

A Discord bot that posts the LeetCode Daily Challenge every morning and walks users through a three-step mock interview (brute force → optimal technique → code) graded by Gemini AI. Scores are tracked on daily, all-time, and practice leaderboards.

Beyond the daily problem, /grind75 and /paretoset pull random problems from curated LeetCode lists so you can drill on demand.

---

Setup

1. Discord Application

1. Go to https://discord.com/developers/applications and create a new application.
2. Bot tab → Create a bot → copy the token → DISCORD_TOKEN.
3. Enable Message Content Intent (required for reading answers in threads/DMs).
4. OAuth2 → URL Generator — scopes: bot, applications.commands.
   Permissions: Send Messages, Embed Links, Add Reactions, Read Message History, Create Public Threads, Send Messages in Threads.
5. Use the generated URL to invite the bot to your server.
6. Enable Developer Mode in Discord (Settings → Advanced), then right-click your server → Copy Server ID → DISCORD_GUILD_ID.
7. Right-click the channel where the daily problem should post → Copy Channel ID → DAILY_CHANNEL_ID.

2. API Keys

• Gemini — https://aistudio.google.com/app/apikey → GEMINI_API_KEY
• Your Discord user ID — Developer Mode → right-click yourself → Copy ID → BOT_OWNER_ID

3. Local Development

cp .env.example .env
# Fill in all values in .env

pip install -r requirements-dev.txt
pytest          # runs all tests (no API keys needed)
python -m leetbot.main

4. Fly.io Deployment

fly launch --no-deploy
fly volumes create leetbot_data --size 1 --region iad
fly secrets set \
  DISCORD_TOKEN=... \
  DISCORD_GUILD_ID=... \
  DAILY_CHANNEL_ID=... \
  GEMINI_API_KEY=... \
  BOT_OWNER_ID=... \
fly deploy

---

Commands

Command	Description
/daily	Repost today's problem embed.
/solve [private:bool]	Start a mock interview on today's problem.
/grind75	Practice a random problem from the Grind 75 list.
/paretoset	Practice a random problem from the Pareto set.
/giveup	Skip the current step, see its solution, and move on.
/explain <question>	Ask a follow-up question about a solution.
/leaderboard daily	Top 10 for today.
/leaderboard alltime	Top 10 all-time by daily points.
/leaderboard practice	Top 10 by practice points.
/stats [user]	Daily and practice points side by side.
/linear	Very important command.
/rate <activity>	Mommy judges what you did today.
/forcedaily (owner)	Force-post today's problem immediately.
/reload (owner)	Reload all cogs without restarting the bot.
/resetattempt [user] [practice] (owner)	Wipe someone's attempt so they can retry.

---

Interview Flow

Each /solve creates a thread off the daily post where only the solving user's messages are routed for grading.

Step	Max Points	Retry Penalty	Floor
Brute Force	15	−3	0
Technique	25	−5	0
Code	60	−10	10

Max total: 100 pts. Use /giveup to lock in partial credit and reveal the reference solution for each individual step.

Grading criteria

The three steps are graded on deliberately different bars:

• Brute force & technique — strict on complexity. You must state both time and space complexity, and both must be correct for the approach you described. Omitting either one is a rejection. Equivalent phrasings all count (O(n), O(N), O(2n), "linear"), and a correct O(n²) for a genuine brute force is correct — you're judged against your own approach, not the optimal one.
• Code — lenient, logic only. Pseudocode is fully acceptable, in any language. Syntax errors, missing imports, missing class Solution wrappers, informal lines like for each x in arr, and unhandled edge cases never cause a rejection. Rejections are reserved for genuinely wrong algorithms or non-attempts. You are not asked to restate complexity here.

---

Practice Mode

/grind75 and /paretoset open a thread, ask which difficulty you want, then serve a random problem you haven't finished yet.

• The difficulty picker only offers difficulties the list actually contains — the Pareto set has no Hard problems, so it shows Easy / Medium / Any.
• 🔁 Different Problem rerolls to another problem at the same difficulty and resets your progress on that thread.
• Grading, hints, /giveup, and the 100-point scale are identical to the daily flow.
• Scores land in a separate practice leaderboard, keyed per problem. Only your personal best per problem is kept, so re-solving something can raise your score but never lower it, and problems appearing in both lists are only ever counted once.

---

Environment Variables

See .env.example for the full list. All variables there are required, the bot fails loudly on startup if any are missing.