client.login("MTIz...") in index.js hands your bot to anyone who can read the repository, and a !say command that runs message.channel.send(args.join(" ")) lets any member make the bot ping @everyone. This skill finds those and the other Discord bot mistakes listed below before you run or ship the bot, and tells you how to fix each one.

Your agent runs a scanner on your discord.js and discord.py projects (including the py-cord, nextcord and disnake forks), HTTP interactions endpoints in Node or Python, and their .env, config, package.json, requirements and README files. It recognises Discord's token and webhook formats (and skips Discord's own documented sample token), follows text from messages, command arguments and slash-command options through assignments, knows that interaction replies and webhooks ping only users by default, and reads the whole project before judging a file: a client-level allowedMentions, a signature check in middleware, an intent used in another file. It reads your .gitignore files, so a token in a config.json or .env that git ignores is not reported as a leak, and it lists at most 50 findings per rule and file. It never runs your bot or contacts Discord. It returns BLOCK, REVIEW, or NOT FLAGGED, with the file, line, reason, and a safer approach. Secrets are never printed.

What it checks

BLOCK (fix before shipping)

• D01: bot tokens, user-account tokens, client secrets and other secret-named values in files git will commit, including literals passed to client.login() or bot.run()
• D02: webhook URLs with their token in files git will commit
• D03: HTTP interactions endpoints and serverless handlers that answer Discord's PING but never call an Ed25519 signature check
• D04: eval, new Function, the vm module, exec or shell commands fed with message text, command arguments or slash-command options, with no check against the bot owner's ID

REVIEW (check before release)

• D11: user text sent as a regular message without allowed-mentions limits or cleaning
• D12: ban, kick, timeout, bulk delete or purge commands without a member permission check
• D13: privileged intents (Message Content, Server Members, Presence, Intents.all()) that no scanned code uses
• D14: messageCreate or on_message handlers that send messages without ignoring bots
• D15: bot invite links, generateInvite or oauth_url asking for Administrator
• D16: tokens or secrets written to console, logger, print or logging calls
• D17: self-bot libraries (discord.js-selfbot-v13, discord.py-self, selfcord, discum) or self_bot=True
• D18: eval or shell commands limited to the bot owner, and the jishaku extension

What's included

• The scanner (Python 3.9+, standard library only)
• A rules guide with the reason, safer approach and official documentation link for each rule
• A risky and a safe example project, and the real report the scanner produces on the risky one

Limits

It reads each file as written and follows chat text through simple assignments in the same file and discord.py command parameters, not through helper functions in other files. Permission checks in custom decorators or middleware with unusual names may not be recognised. Intent use is judged from the scanned files, so scan the whole project. Command permissions changed by server admins, Developer Portal settings and the bot's real role permissions are not visible in code, and git history is not read. Libraries other than discord.js, discord.py and its forks (such as Eris, hikari, discordgo or JDA) are not checked. NOT FLAGGED is not a guarantee.

Contents

One ZIP file with 25 files: SKILL.md, the scanner in scripts/ (6 Python files), references/rules.md (every rule with its reason and safer approach), and examples/ (16 example files plus report.md, the real scanner output on the risky example).

How to use: This is an agent skill: a folder with a SKILL.md file and Python scripts that an AI coding agent such as Claude Code, Cursor, or Codex CLI can run. Unzip it into your agent's skills folder (for Claude Code: ~/.claude/skills/), then ask your agent to check your project. The scripts also run on their own with Python 3.9 or newer, with nothing to install. They read your files only: no network access, and nothing is sent anywhere.