HTB Kobold MCPJam RCE, PrivateBin LFI-to-RCE, and Docker Hos...

[carlospolop]

🤖 Automated Content Update

This PR was automatically generated by the HackTricks News Bot based on a technical blog post.

📝 Source Information

• Blog URL: https://0xdf.gitlab.io/2026/08/01/htb-kobold.html
• Blog Title: HTB Kobold: MCPJam RCE, PrivateBin LFI-to-RCE, and Docker Host Takeover
• Suggested Section: AI Security > AI MCP Security for MCPJam unauthenticated command execution; Pentesting Web > File Inclusion/Path Traversal for PrivateBin template-cookie LFI and LFI-to-RCE; Linux Hardening > Containers, Namespaces > Container Security > Sensitive Host Mounts for writable bind-mount abuse and Docker host takeover

🎯 Content Summary

Overview

Kobold is an Easy-rated Linux Hack The Box machine exposing OpenSSH, Nginx virtual hosts, MCPJam Inspector, PrivateBin, and the Arcane Docker management panel. The compromise chains two unauthenticated vulnerabilities with an unsafe Docker bind mount and credential reuse. CVE-2026-23744 executes attacker-controlled processes through MCPJam and provides a host shell as ben. CVE-2025-64714 traverses PrivateBin's template path and i...

🔧 Technical Details

Exposed development APIs: Development and inspector services may legitimately accept executable names, argument arrays, and environment variables to launch local subprocesses. If such an API listens on all interfaces without authentication, endpoints that create, connect to, or start servers may provide remote command execution. Verify execution with an ICMP, DNS, or HTTP callback before attempting a shell. A protocol-level connection error does not necessarily mean the operating-system process failed to execute.

Template-cookie LFI: Applications may store the selected theme or template in a cookie and later use it in a filesystem include. Confirm that template selection is enabled by observing the interface and resulting cookies. Test traversal using a known file outside the templates directory. An HTTP 500 response can indicate successful inclusion when the selected file is valid PHP but not a valid application template.

LFI-to...

🤖 Agent Actions

Updated src/AI/AI-MCP-Servers.md.

Added a new subsection on remote process-launch APIs exposed beyond loopback covering:

• inspector/dev endpoints that spawn MCP servers from client-supplied command / args / env
• why public or unauthenticated exposure becomes RCE
• why MCP/protocol errors can still mean the OS command executed
• callback-based verification guidance
• auditing and hardening points

Also updated the file’s References section with:

• the MCPJam advisory
• the HTB Kobold blog post

I chose the AI MCP page because the MCPJam technique was the most clearly missing, while the PrivateBin/LFI and Docker host-mount material is already substantially covered elsewhere in HackTricks.

---

This PR was automatically created by the HackTricks Feed Bot. Please review the changes carefully before merging.

[carlospolop]

🔗 Additional Context

Original Blog Post: https://0xdf.gitlab.io/2026/08/01/htb-kobold.html

Content Categories: Based on the analysis, this content was categorized under "AI Security > AI MCP Security for MCPJam unauthenticated command execution; Pentesting Web > File Inclusion/Path Traversal for PrivateBin template-cookie LFI and LFI-to-RCE; Linux Hardening > Containers, Namespaces > Container Security > Sensitive Host Mounts for writable bind-mount abuse and Docker host takeover".

Repository Maintenance:

• MD Files Formatting: 990 files processed

Review Notes:

• This content was automatically processed and may require human review for accuracy
• Check that the placement within the repository structure is appropriate
• Verify that all technical details are correct and up-to-date
• All .md files have been checked for proper formatting (headers, includes, etc.)

Bot Version: HackTricks News Bot v1.0