CVE-2024-21663: Discord-recon Bot RCE Vulnerability

CVE-2024-21663 Overview

CVE-2024-21663 is a command injection vulnerability in Discord-Recon, a Discord bot developed by demon1a for automating bug bounty reconnaissance, scans, and information gathering. The flaw lets any authenticated Discord user execute arbitrary shell commands on the host running the bot, without requiring an admin role. The issue stems from unsanitized user input passed directly to shell commands through subprocess.Popen with shell=True. The maintainer fixed the vulnerability in version 0.0.8 by introducing a CommandInjection.sanitizeInput routine applied to command arguments.

Critical Impact
Any non-admin Discord user in a server running Discord-Recon can execute arbitrary shell commands on the host, resulting in full compromise of the bot's runtime environment.

Affected Products

demon1a/discord-recon versions prior to 0.0.8
demon1a/discord-recon0.0.8:beta release channel
Any Discord server hosting a vulnerable Discord-Recon bot instance

Discovery Timeline

2024-01-09 - CVE-2024-21663 published to NVD
2026-06-17 - Last updated in NVD database

Technical Details for CVE-2024-21663

Vulnerability Analysis

Discord-Recon exposes several bot commands (for example, prips) that accept user-supplied arguments and pass them into shell subprocesses. The bot uses subprocess.Popen with shell=True, which invokes /bin/sh -c and interprets shell metacharacters such as ;, &&, |, and backticks. Without input sanitization, an attacker can append arbitrary shell commands to the argument string. The vulnerability is classified under CWE-77 (Command Injection) and CWE-20 (Improper Input Validation). Exploitation requires only low-privileged access, meaning any user in the Discord server can trigger the flaw. Successful exploitation grants command execution under the bot's operating system account, allowing reconnaissance data theft, lateral movement, or persistent implants.

Root Cause

The root cause is direct string interpolation of Discord-supplied arguments into a shell command line. The vulnerable pattern is subprocess.Popen(f"prips {argument}", shell=True) where argument originates from the Discord message. No allow-listing, escaping, or shell disabling was applied before the fix.

Attack Vector

An attacker with a standard (non-admin) role in a Discord server that hosts the bot sends a bot command message. By appending shell metacharacters to the expected input, the attacker chains commands executed by the underlying shell. The attack is remote, low-complexity, and requires no user interaction beyond the attacker sending a message.

python

# Security patch in app.py (commit f9cb0f6)
@Client.command()
async def prips(ctx, *, argument):
    argument = CommandInjection.sanitizeInput(argument)
    Output = subprocess.Popen(f"prips {argument}", stdout=subprocess.PIPE, stderr=subprocess.STDOUT, shell=True)
    Output = Output.communicate()[0].decode('UTF-8')
# Source: https://github.com/DEMON1A/Discord-Recon/commit/f9cb0f67177f5e2f1022295ca8e641e47837ec7a
# The fix wraps user-controlled `argument` in CommandInjection.sanitizeInput before shell execution.

Detection Methods for CVE-2024-21663

Indicators of Compromise

Unexpected child processes spawned by the Discord-Recon Python process, such as sh, bash, curl, wget, nc, or python -c.
Outbound network connections from the bot host to unfamiliar destinations following bot command invocations.
Discord bot logs showing command arguments containing shell metacharacters like ;, &&, |, $(), or backticks.
New files, cron jobs, or systemd units created under the bot's service account after a bot command was processed.

Detection Strategies

Monitor process ancestry for python or python3 processes spawning shell interpreters or reconnaissance tooling not part of the bot's normal command set.
Alert on Discord-Recon commands whose arguments contain shell control characters or IFS-sensitive tokens.
Correlate bot command execution timestamps with outbound network activity from the host to identify data exfiltration.

Monitoring Recommendations

Enable verbose logging for the bot process and forward logs to a centralized log platform for retention and search.
Track filesystem writes and privilege-changing syscalls under the bot's UID using auditd or eBPF-based sensors.
Baseline expected outbound destinations for the bot host and alert on deviations.

How to Mitigate CVE-2024-21663

Immediate Actions Required

Upgrade Discord-Recon to version 0.0.8 or later, which introduces CommandInjection.sanitizeInput on user-controlled arguments.
Restrict which Discord roles can invoke bot commands until the upgrade is confirmed in production.
Review bot host logs and process history for prior exploitation and rotate any credentials accessible from the host.

Patch Information

The fix is available in the GitHub commit f9cb0f6 and documented in GHSA-fjcj-g7x8-4rp7. Additional context is in GitHub Issue #23. Upgrade to version 0.0.8 to remediate.

Workarounds

Take the bot offline until the upgrade to 0.0.8 is applied if immediate patching is not possible.
Run the bot inside a hardened container with a read-only filesystem, dropped capabilities, and no outbound egress except to Discord APIs.
Restrict Discord server membership so that only trusted users can send messages to channels the bot listens on.

bash

# Configuration example: upgrade and containment
pip install --upgrade discord-recon==0.0.8

# Example hardened systemd unit for the bot
# /etc/systemd/system/discord-recon.service
[Service]
User=discord-recon
NoNewPrivileges=true
ProtectSystem=strict
ProtectHome=true
PrivateTmp=true
CapabilityBoundingSet=
RestrictSUIDSGID=true
ExecStart=/usr/bin/python3 /opt/discord-recon/app.py