Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@pentool-mcp-server find non-obvious endpoints on example.com"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
English · Русский (RU)
MCP server for Pentool — the local AI assistant for web pentesting
A self-contained stdio JSON-RPC 2.0 server that brings AI capabilities to any MCP client — first of all to Pentool (BYO-LLM: picking checks, bypassing WAF, finding non-obvious endpoints). Installed with uv in one command.
MCP is the open protocol for connecting LLM
models to tools. pentool-mcp-server encapsulates Pentool's MCP layer as a
reusable PyPI package: attached locally (stdio), no network, no ports, installs
in seconds, no heavy dependencies.
Related MCP server: pentestMCP
This server is the MCP component of Pentool, a professional web-security testing terminal (Burp-compatible proxy, scanner, spider, intruder — all in a TUI). It powers Pentool's AI:
🎯 picking the relevant scan checks for a concrete target;
🛡 WAF bypass / payload suggestions;
🕷 finding non-obvious endpoints during spider crawling.
Try the full stack:
uv tool install "pentool[ai]"→pentool ai setup→pentool.
# Install uv (if not present): https://docs.astral.sh/uv/
curl -LsSf https://astral.sh/uv/install.sh | sh
# Standalone — just the MCP server
uv tool install pentool-mcp-server
pentool-mcp-server --version
# Or together with Pentool and its AI extras
uv tool install "pentool[ai]"Health check (stdio):
echo '{"jsonrpc":"2.0","id":1,"method":"ping"}' | pentool-mcp-server
# → {"jsonrpc": "2.0", "id": 1, "result": {"status": "ok"}}uv is the standard install path — isolated environment, exactly how Pentool itself is installed. It keeps the server out of your system Python.
The server opens no ports and runs as a local subprocess — safe on any host. With uv it stays out of the system Python:
uv tool install pentool-mcp-serverFull user guide — docs/GUIDE.md. Russian — README.ru.
Tool | Description |
| MCP protocol handshake |
| List available tools |
| Generate an LLM answer for a task ( |
| Readiness check (is a model installed) |
| Point to a GGUF model path |
| Process liveness |
No network by default. pentool-mcp-server listens only on stdio
(stdin → stdout) inside the local process. External access is impossible —
neither from other processes nor from another host.
Do not expose it over TCP/0.0.0.0 without auth. If run on a network port
outwardly, anyone able to write to stdin/port gets tools/call generate
(resource usage + sending target data to the LLM). MCP has no built-in
authentication — bind to 127.0.0.1 and gate via a firewall.
Target privacy. Data (URL, payload) leaves the host only if an external LLM provider is connected; with a local GGUF model the traffic stays on the machine.
More — docs/GUIDE.md.
stdio JSON-RPC 2.0 server (initialize / tools/list / tools/call / ping)
generate, health, configure tools
real LLM runner (llama-cpp-python) in model.py
optional TCP mode with HMAC auth
AGPL-3.0 — the same license as Pentool.
Support / bugs: issues · Designed for Pentool · User Guide
Принять параметры конфигурации сервера (напр. путь модели).
Inputs: model.
Сгенерировать ответ LLM по задаче. payload — строка запроса (данные цели), model — путь к GGUF-файлу модели (необязательно). При отсутствии модели вернёт ошибку setup required.
Inputs: task (required), payload (required), model.
Проверка готовности: вернёт статус, установлена ли модель.
Inputs: no named parameters.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 3 tools
Each tool has a clearly distinct purpose: 'generate' for LLM inference, 'health' for readiness checking, and 'configure' for setting parameters. There is no overlap or ambiguity between them.
All tool names are simple, lowercase verbs that accurately describe their action ('generate', 'health', 'configure'). The pattern is uniform and predictable.
With 3 tools, the count is appropriate for a minimal MCP server focused on LLM inference. It could potentially benefit from a status tool or a list-models tool, but the current set is sufficient for its core purpose.
The tool surface covers basic lifecycle: configure (setup), health (readiness), and generate (infer). However, there is no tool for cancelling generation, listing available models, or inspecting current configuration, which are notable gaps for a production server.
MCP server for Pentest-Tools.com: run scans, manage findings and reports via your preffered LLM.
Security scanner for MCP servers. Detect vulnerabilities, prompt injection, and tool poisoning.
Security scanner for AI agents and MCP servers: injection, jailbreak, tool abuse.
Security-first WordPress MCP server. 129 tools for Claude, ChatGPT, Gemini. Free on wp.org.
