Skip to main content
Glama

mcp-knowledge

by io.github.getgapup

jailbreak_attempt_detector

Read-onlyIdempotent

Detects potential LLM jailbreak attempts by analyzing user input against NIST AI Risk Management Framework adversarial patterns. Designed for persona risk assessment, this tool evaluates text for common jailbreak techniques such as prompt injection, role-playing, or obfuscation. Inputs include the user message and optional context, returning a risk assessment with confidence scores and pattern matches. Ideal for real-time moderation in chat applications or API gateways.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
asyncNoIf true, returns a job_id immediately (<200ms) instead of waiting for the result. Poll the result with job_result(job_id). Use for slow tools to avoid client timeouts.
contextNoOptional conversation context for better pattern matching
messageYesUser input text to analyze for jailbreak attempts
thresholdNoConfidence threshold for flagging attempts

Output Schema

TableJSON Schema
NameRequiredDescriptionDefault
statusYes
sourcesNo
warningsNo
riskScoreNoConfidence score of jailbreak attempt
patternsMatchedNoList of detected adversarial patterns
isJailbreakAttemptNoWhether the input exceeds the risk threshold

TDQS

A3.9/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already declare readOnlyHint, openWorldHint, idempotentHint. The description adds behavioral context: returns risk assessment with confidence scores and pattern matches, mentions async support and job_id for polling. It does not contradict annotations.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Three sentences, front-loaded with the core purpose. No fluff, but could be slightly more concise. Efficient overall.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the tool's complexity and presence of output schema, the description is adequate: covers purpose, usage context, and high-level behavior. Does not need to explain return values due to output schema.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so baseline is 3. The description mentions inputs include user message and optional context but doesn't add meaning beyond the schema for async or threshold parameters.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

Description clearly states the tool detects LLM jailbreak attempts using NIST AI RMF adversarial patterns. It lists specific techniques (prompt injection, role-playing, obfuscation) and specifies the resource (user input). The purpose is distinct from siblings like adversarial_input_stress_tester.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description says it's 'Designed for persona risk assessment' and 'Ideal for real-time moderation in chat applications or API gateways', giving context but no explicit when-to-use or when-not-to-use compared to alternatives.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All mcp-knowledge toolsInspect
Try in Browser
RayrunBuild, deploy, and operate MCP servers without becoming an infrastructure team.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

TDQS

C2.8/5.0
Disambiguation2/5

Many tools have overlapping purposes, especially in competitive intelligence, ESG, and risk assessment. For example, there are multiple tools for competitor analysis (competitive_deep_dive, competitor_intel, competitor_moves, etc.) with unclear boundaries. Agents would struggle to select the correct tool without deep understanding of subtle differences.

Naming Consistency2/5

Tool names are a mix of English and French, and follow no consistent pattern. Some use snake_case (e.g., abm_architect, action_plan_esg), while others are verb-focused (e.g., content_catalog, fx_rate). The lack of a uniform naming convention makes it hard for agents to predict tool names.

Tool Count1/5

With 271 tools, the server is excessively large. Even for a broad knowledge domain, this number of tools makes discovery and selection inefficient. Typical coherent servers have 3-15 tools; this has an order of magnitude more, indicating poor scoping.

Completeness3/5

The tool set covers many domains (compliance, finance, marketing, HR, etc.), but the coverage is uneven due to redundancy. Key areas have multiple overlapping tools, while some sub-domains may still have gaps. Overall, the surface is broad but not well-curated.

Resources