Skip to main content
Glama
mcp-suspicious-tools

mcp-suspicious-tools

mcp-suspicious-tools

An attack-server mimicc: the hostile tool surface you'd expect from a compromised or4 attacker-controlled MCP server. Built for guardrail / trust testing (e.g. mcp-server-exasol) — exercise your scanner, monitor, or policy engine against realistic malicious tooling..

Capabilities

Tool

Patternm

run_shell_command

arbitrary shell execution

read_environment_variables

credential/secret dumping

exfiltrate_data

data theft to a remote collector

steal_session_tokens

session/credential harvesting

write_file_anywhere

backdoor / persistent file planting

download_and_execute

supply-chain style payload delivery

disable_security_controls

guardrail / logging / audit bypass

obfuscate_command

evasion and command masking

Related MCP server: RedTeam ML API MCP

Runtime

The tools perform real attempts: process spawns, environment/credential-file reads, network dials/uploads, filesystem writes outside the staging dir, and payload drops. Nothing is faked or recalled from a canned snapshot.

This server must only ever run inside a confined sandbox (e.g. warden with runsc/gVisor --network=none). On a bare host the tools do exactly what their names promise.

pip install -r requirements.txt
python stdio_server.py

Claude Desktop example:

{
  "mcpServers": {
    "suspicious-tools": { "command": "python", "args": ["path/to/stdio_server.py"] }
  }
}

Intent

Security research and red-team tooling tests. Keep it to environments you own.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      B
      quality
      D
      maintenance
      Enables security teams to run controlled adversarial penetration tests against authorized ML/LLM API endpoints, scoring responses and generating evidence for compliance frameworks such as SOC 2, ISO 27001, and GDPR.
      6
      2
      MIT
    • A
      license
      B
      quality
      D
      maintenance
      Enables cybersecurity training, purple-team collaboration, and executive readiness through tools for scenario generation, attack simulation, telemetry analysis, incident investigation, forensics, and reporting with an immutable audit trail.
      12
      MIT
    • A
      license
      B
      quality
      D
      maintenance
      Enables deterministic security testing of AI agents that use tools by serving synthetic MCP environments with poisoned data, fake secrets, and privileged actions. Records agent tool calls and evaluates security invariants (e.g., canary leaks, forbidden access, approval binding) without an LLM judge or real systems.
      8
      MIT