Cybersec Toolkit
Authorization-gated MCP server to run 670+ security tools for CTF, pentest and DFIR
- Pricing
- Open source
- Platforms
- Linux, CLI, self-hosted
- Runs
- Locally
- Version
- 1.5.0
- Listing updated
- today
Similar MCP servers
About Cybersec Toolkit
Cybersec Toolkit is an open-source MCP server for security. It runs locally on your machine and is started by your AI client.
The default is a sandbox. Tools launched through the server run inside a disposable Kata Containers VM with its own kernel, no Docker socket and dropped capabilities, and startup fails closed instead of falling back to the host. A host mode exists for machines without KVM, such as macOS or Windows, and bubblewrap can wrap tools in that mode.
The server shares one tool registry with a bash installer that sets up the toolkit on Debian, Ubuntu, Kali, Parrot, Fedora, Arch, openSUSE and Termux. The README lists 15 MCP tools for discovery, advice and governed execution. Execution goes through an allowlist, argument checks, a per-tool list of blocked flags, network policy, rate limits and output caps, without invoking a shell. Calls are recorded in a hash-chained audit log.
External targets and arbitrary script execution are off by default and need environment variables to be set. A scope file can turn the VM into a default-deny network boundary for an engagement. The repository also bundles 872 Agent Skills for CTF, pentest, bug bounty and DFIR work.
Install
Run in a terminal
claude mcp add cybersec-toolkit -- docker run -i --rm ghcr.io/26zl/cybersec-toolkit:1.5.0Add to claude_desktop_config.json
{
"mcpServers": {
"cybersec-toolkit": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"ghcr.io/26zl/cybersec-toolkit:1.5.0"
]
}
}
}Add to ~/.cursor/mcp.json
{
"mcpServers": {
"cybersec-toolkit": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"ghcr.io/26zl/cybersec-toolkit:1.5.0"
]
}
}
}Add to .vscode/mcp.json
{
"servers": {
"cybersec-toolkit": {
"type": "stdio",
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"ghcr.io/26zl/cybersec-toolkit:1.5.0"
]
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"cybersec-toolkit": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"ghcr.io/26zl/cybersec-toolkit:1.5.0"
]
}
}
}Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"cybersec-toolkit": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"ghcr.io/26zl/cybersec-toolkit:1.5.0"
]
}
}
}Using several servers? Build one config file with the MCP config generator.
Configuration
CYBERSEC_MCP_ALLOW_EXTERNALSet to 1 only for explicitly authorized external scopes.CYBERSEC_MCP_ALLOW_SCRIPTSSet to 1 to enable unsandboxed run_script execution.
Tools this server provides (3)
Taken from the project documentation; the running server may differ.
- run_tool
- Runs a registered security tool through the governed execution path.
- run_pipeline
- Runs a chain of registered tools under the same governance.
- run_script
- Runs arbitrary code; disabled unless explicitly enabled.
What it does
- Governed tool execution
- Runs registered tools through an allowlist, argument sanitisation and network policy, with no shell.
- Disposable Kata VM
- Each session gets its own VM that is destroyed when the client disconnects.
- Scope file firewall
- A listed set of addresses and ports becomes a default-deny network boundary inside the VM.
- Hash-chained audit log
- Every call is recorded in a tamper-evident log on the host.
- Opt-in risky features
- External targets and run_script stay disabled until you set the matching variables.
- Agent Skills bundle
- 872 skills supply methodology for CTF, pentest, bug bounty and DFIR.
Good for
- Let an AI assistant triage a binary in a CTF
- Map the attack surface of a lab network you own
- Support incident response and DFIR work with scripted tooling
- Keep an audit trail of what an agent ran during an engagement
Pricing
Open source
Open-source repository; no hosted plan mentioned
Cybersec Toolkit FAQ
Does it run tools on my machine?
By default in a Kata Containers VM; the --local flag runs the server on the host instead.
Can it scan external targets?
Not by default. Targets that do not resolve to private or loopback ranges are rejected until CYBERSEC_MCP_ALLOW_EXTERNAL is set to 1.
Can the agent run arbitrary scripts?
Only if you set CYBERSEC_MCP_ALLOW_SCRIPTS to 1, which enables run_script. It is unsandboxed code execution.
What does the sandbox need?
Linux with KVM, Docker 23+ with a Kata runtime and Node.js 22+.
Is Cybersec Toolkit free?
Yes. Cybersec Toolkit is open source, so you can use and self-host it for free.
How do I add Cybersec Toolkit to Claude Code?
Run this command in your terminal: claude mcp add cybersec-toolkit -- docker run -i --rm ghcr.io/26zl/cybersec-toolkit:1.5.0
What are the best alternatives to Cybersec Toolkit?
The closest alternatives to Cybersec Toolkit are REA, AiSOC, redamon, x64dbg-mcp-server and Agent Roast.
Where this comes from
- Description: written from the project's README.
- Install steps: from the official MCP Registry entry.



