Skip to main content
Glama
EdenYavin

Garak-MCP

by EdenYavin
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Hybrid

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault

No arguments

Instructions

Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.

This server publishes no instructions, or was last inspected before Glama recorded them.

Capabilities

Features and capabilities supported by this server

Protocol revision2025-11-25

CapabilityDetails
tools
{
  "listChanged": false
}
prompts
{
  "listChanged": false
}
resources
{
  "subscribe": false,
  "listChanged": false
}
experimental
{}

Tools

Functions exposed to the LLM to take actions

NameDescription
list_model_typesB
List all available model types.

Returns:
    list[str]: A list of available model types.
list_modelsB
List all available models for a given model type.
Those models can be used for the attack and target models.

Args:
    model_type (str): The type of model to list (ollama, openai, huggingface, ggml)

Returns:
    list[str]: A list of available models.
list_garak_probesB
List all available Garak attacks.

Returns:
    list: A list of available probes / attacks.
get_reportC
Get the report of the last run.

Returns:
    str: The path to the report file.
run_attackC
Run an attack with the given model and probe which is a Garak attack.

Args:
    model_type (str): The type of model to use.
    model_name (str): The name of the model to use.
    probe_name (str): The name of the attack / probe to use.

Returns:
    list: A list of vulnerabilities.

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription

No resources

Deploy Server
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

B3.4/5.0

Scored across 5 tools

Disambiguation5/5

Each tool has a clearly distinct purpose: get_report retrieves results, list_garak_probes lists attacks, list_models lists models by type, list_model_types lists types, and run_attack executes an attack. There is no overlap in functionality, making tool selection unambiguous.

Naming Consistency5/5

All tool names follow a consistent verb_noun pattern (e.g., get_report, list_garak_probes, run_attack). The naming is uniform across all tools, with no deviations in style or convention.

Tool Count5/5

With 5 tools, the set is well-scoped for a Garak MCP server focused on AI security testing. Each tool serves a specific role in the workflow, from setup (listing models/probes) to execution (run_attack) and review (get_report), without unnecessary redundancy.

Completeness4/5

The tool surface covers the core AI security testing workflow: listing resources (models, probes, types), running attacks, and retrieving reports. A minor gap exists in lacking tools for configuring attack parameters or managing attack history, but agents can still perform essential tasks effectively.

Maintenance

ActivityInactive
ResponsivenessNo issues

Resources