Systemores AI Agent & MCP Security Auditor

AI agent security, MCP 2026 schema & graph audits, <250ms guardrails, and compliance dossiers

Should I use this

Quality & Safety

A
Description quality
97%
Schema completeness
100%
Naming quality
75%
Poisoning risk
100%
Permission match
100%
Protocol compliance
100%

Findings (3)

  • LOWTool 'systemores_audit_tool_integrity' name length outside 3-30 rangein systemores_audit_tool_integrity
  • LOWTool 'systemores_inject_environment_fault' name length outside 3-30 rangein systemores_inject_environment_fault
  • LOWTool 'systemores_analyze_trace_telemetry' name length outside 3-30 rangein systemores_analyze_trace_telemetry

Based on automated analysis of tool definitions and protocol compliance.

Context Cost

~4,286Tokens (tool definitions)
~3.3 KBTypical response size
Significant attention impact (3.35% of 128k context)

This is the approximate number of tokens consumed each time the server's tools are loaded into a model's context. Higher counts reduce the attention available for other tasks.

Install

One-Click Install

Add this to your `claude_desktop_config.json` file:

{
  "mcpServers": {
    "systemores-plugin": {
      "command": "npx",
      "args": [
        "@systemores/systemores-plugin"
      ]
    }
  }
}

Runnable packages

npm@systemores/systemores-plugin1.2.0stdio

Remote endpoints

https://audit.systemores.com/api/mcpstreamable-http

What it can do

Tool inventory

Tools (12)

🟢 Read-only🟡 Write🔴 Delete⚪ Unknown
🟢systemores_stress_test_agent(systemPrompt, role, pressIntensity)

Exposes the target agent persona to a battery of adversarial prompts (Presses) and evaluates behavioral stability using Murray's Needs-Press diagnostic model.

Input Schema

{
  "type": "object",
  "properties": {
    "systemPrompt": {
      "type": "string",
      "description": "The behavioral blueprint or internal operational instructions of the target agent. Maximum 8,000 characters."
    },
    "role": {
      "type": "string",
      "description": "The intended operational function of the agent.",
      "enum": [
        "customer_support",
        "financial_underwriter",
        "medical_assistant",
        "general_developer"
      ]
    },
    "pressIntensity": {
      "type": "string",
      "description": "The intensity level of adversarial testing. 'high' simulates prompt injection batteries.",
      "enum": [
        "low",
        "medium",
        "high"
      ]
    }
  },
  "required": [
    "systemPrompt",
    "role"
  ]
}

Output Schema

{
  "type": "object",
  "properties": {
    "status": {
      "type": "string",
      "enum": [
        "PASSED",
        "FAILED"
      ]
    },
    "metrics": {
      "type": "object"
    },
    "attacksSimulated": {
      "type": "integer"
    },
    "breachesIdentified": {
      "type": "array"
    },
    "recommendation": {
      "type": "string"
    },
    "derivedPsychometrics": {
      "type": "object"
    }
  },
  "required": [
    "status",
    "metrics",
    "attacksSimulated",
    "breachesIdentified",
    "recommendation"
  ]
}
🟢systemores_evaluate_drift(systemPrompt, turns, temperature)

Simulates multiple conversational turns to evaluate agent output consistency and behavioral drift over extended interactions.

Input Schema

{
  "type": "object",
  "properties": {
    "systemPrompt": {
      "type": "string",
      "description": "The target agent's system prompt. Maximum 8,000 characters."
    },
    "turns": {
      "type": "integer",
      "description": "The number of conversational turns to simulate (1–100).",
      "default": 10,
      "minimum": 1,
      "maximum": 100
    },
    "temperature": {
      "type": "number",
      "description": "The LLM temperature parameter for the target agent (0.0–1.5). Higher values increase drift probability.",
      "default": 0.7,
      "minimum": 0,
      "maximum": 1.5
    }
  },
  "required": [
    "systemPrompt"
  ]
}

Output Schema

{
  "type": "object",
  "properties": {
    "turnsEvaluated": {
      "type": "integer"
    },
    "averageDriftIndex": {
      "type": "number"
    },
    "entropyRating": {
      "type": "string",
      "enum": [
        "HIGH",
        "CONTROLLED"
      ]
    },
    "failurePointDetected": {
      "type": "string"
    },
    "estimatedDeviationAngleDegree": {
      "type": "integer"
    },
    "derivedPsychometrics": {
      "type": "object"
    }
  },
  "required": [
    "turnsEvaluated",
    "averageDriftIndex",
    "entropyRating",
    "failurePointDetected",
    "estimatedDeviationAngleDegree"
  ]
}
🟢systemores_assess_bias(systemPrompt, complianceDomain)

Verifies the agent's compliance with fair-lending, fair-housing, employment non-discrimination (EEOC/SIOP/ECOA), child-directed safety (COPPA), or EU AI Act (EU 2024/1689) standards.

Input Schema

{
  "type": "object",
  "properties": {
    "systemPrompt": {
      "type": "string",
      "description": "The underwriting, hiring, housing, or EU AI Act high-risk agent system prompt to evaluate. Maximum 8,000 characters."
    },
    "complianceDomain": {
      "type": "string",
      "description": "The legal compliance framework to evaluate against (employment, housing, lending, COPPA child safety, or EU AI Act 2024/1689).",
      "enum": [
        "employment_eeoc",
        "housing_fair",
        "lending_ecoa",
        "children_coppa",
        "eu_ai_act"
      ]
    }
  },
  "required": [
    "systemPrompt",
    "complianceDomain"
  ]
}

Output Schema

{
  "type": "object",
  "properties": {
    "framework": {
      "type": "string"
    },
    "riskRating": {
      "type": "string",
      "enum": [
        "LOW",
        "MEDIUM",
        "HIGH"
      ]
    },
    "nonComplianceFlags": {
      "type": "array"
    },
    "mitigationInstructions": {
      "type": "string"
    },
    "derivedPsychometrics": {
      "type": "object"
    }
  },
  "required": [
    "framework",
    "riskRating",
    "nonComplianceFlags",
    "mitigationInstructions"
  ]
}
🟢systemores_measure_drag(promptLength, responseLength, latencyMs)

Measures operational latency, response length, and execution drag parameters to analyze cognitive throughput stability within the Behavioral OS ecosystem.

Input Schema

{
  "type": "object",
  "properties": {
    "promptLength": {
      "type": "integer",
      "description": "The character length of the prompt sent to the agent.",
      "minimum": 0
    },
    "responseLength": {
      "type": "integer",
      "description": "The character length of the agent's response.",
      "minimum": 0
    },
    "latencyMs": {
      "type": "number",
      "description": "The end-to-end response latency in milliseconds. Must be >= 1.",
      "minimum": 1
    }
  },
  "required": [
    "promptLength",
    "responseLength",
    "latencyMs"
  ]
}

Output Schema

{
  "type": "object",
  "properties": {
    "characterPerMs": {
      "type": "number"
    },
    "dragRating": {
      "type": "string",
      "enum": [
        "EFFICIENT",
        "HIGH"
      ]
    },
    "estimatedTokenCostRating": {
      "type": "string",
      "enum": [
        "OPTIMIZED",
        "EXPENSIVE"
      ]
    },
    "cognitiveEfficiencyScore": {
      "type": "number"
    }
  },
  "required": [
    "characterPerMs",
    "dragRating",
    "estimatedTokenCostRating",
    "cognitiveEfficiencyScore"
  ]
}
🟢systemores_tool_integrity(systemPrompt, toolContext)

Evaluates whether the agent's system prompt establishes the behavioral conditions for safe, reliable tool use — auditing tool selection discipline, argument construction fidelity, and failure resilience when tool returns are ambiguous or erroneous.

Input Schema

{
  "type": "object",
  "properties": {
    "systemPrompt": {
      "type": "string",
      "description": "The agent's system prompt to evaluate for tool use safety. Maximum 8,000 characters."
    },
    "toolContext": {
      "type": "string",
      "description": "The domain of tool usage being audited. Affects risk weighting.",
      "enum": [
        "api_calls",
        "database",
        "filesystem",
        "code_execution"
      ]
    }
  },
  "required": [
    "systemPrompt"
  ]
}

Output Schema

{
  "type": "object",
  "properties": {
    "status": {
      "type": "string",
      "enum": [
        "PASSED",
        "FAILED"
      ]
    },
    "toolContext": {
      "type": "string"
    },
    "metrics": {
      "type": "object"
    },
    "risks": {
      "type": "array"
    },
    "recommendation": {
      "type": "string"
    }
  },
  "required": [
    "status",
    "toolContext",
    "metrics",
    "risks",
    "recommendation"
  ]
}
🟢systemores_audit_tool_integrity(toolDefinitionsJson, strictMode)

Audits the structural integrity, naming conventions, schema completeness, safety annotations, and injection surface of an AI agent's exposed tool definitions. Identifies malformed schemas, privilege overreach, deceptive safety hints, duplicate names, and description quality gaps. Operates entirely offline — no external API calls required.

Input Schema

{
  "type": "object",
  "properties": {
    "toolDefinitionsJson": {
      "type": "string",
      "description": "A JSON-serialized array of tool definition objects to audit. Each object should contain at minimum 'name' and 'description' fields. Accepts standard MCP ListToolsResult format."
    },
    "strictMode": {
      "type": "boolean",
      "description": "When true, applies stricter scoring thresholds — missing optional fields such as parameter descriptions, inputSchema, and MCP tool annotations are penalized. Defaults to false.",
      "default": false
    }
  },
  "required": [
    "toolDefinitionsJson"
  ]
}

Output Schema

{
  "type": "object",
  "properties": {
    "overallScore": {
      "type": "number"
    },
    "integrityRating": {
      "type": "string",
      "enum": [
        "PASS",
        "REVIEW",
        "FAIL"
      ]
    },
    "toolsAudited": {
      "type": "integer"
    },
    "findings": {
      "type": "array"
    },
    "summary": {
      "type": "string"
    }
  },
  "required": [
    "overallScore",
    "integrityRating",
    "toolsAudited",
    "findings",
    "summary"
  ]
}
🟢systemores_audit_agent_graph(graphTopologyJson, maxDepth, strictPrivilegeCheck)

Statically scans multi-agent execution topologies (DAGs/meshes) using DFS algorithms to detect cyclic delegation loops and privilege escalation paths from public untrusted sub-agents to backend write/admin tools.

Input Schema

{
  "type": "object",
  "properties": {
    "graphTopologyJson": {
      "type": "string",
      "description": "JSON-serialized graph topology object containing 'nodes' (id, privilegeLevel) and 'edges' (source, target)."
    },
    "maxDepth": {
      "type": "integer",
      "description": "Maximum depth to traverse during cycle and privilege path search.",
      "default": 10
    },
    "strictPrivilegeCheck": {
      "type": "boolean",
      "description": "When true, flags any path from 'public'/'user' nodes to 'admin'/'system' nodes as a breach.",
      "default": true
    }
  },
  "required": [
    "graphTopologyJson"
  ]
}

Output Schema

{
  "type": "object",
  "properties": {
    "status": {
      "type": "string",
      "enum": [
        "PASSED",
        "FAILED"
      ]
    },
    "metrics": {
      "type": "object"
    },
    "cyclesDetected": {
      "type": "array"
    },
    "privilegeEscalationPaths": {
      "type": "array"
    },
    "risks": {
      "type": "array"
    },
    "recommendation": {
      "type": "string"
    }
  },
  "required": [
    "status",
    "metrics",
    "cyclesDetected",
    "privilegeEscalationPaths",
    "risks",
    "recommendation"
  ]
}
🟢systemores_generate_cohort(domain, cohortSize, includeAdversarial, locale)

Generates statistical cohorts of synthetic user personas across enterprise operational domains with IPIP-NEO-120 30-facet psychometric profiles, demographic diversity, and prompt injection payloads.

Input Schema

{
  "type": "object",
  "properties": {
    "domain": {
      "type": "string",
      "description": "The operational domain for persona generation.",
      "enum": [
        "underwriting",
        "customer_support",
        "healthcare",
        "devops",
        "claims"
      ]
    },
    "cohortSize": {
      "type": "integer",
      "description": "Number of synthetic personas to generate (1–100).",
      "default": 10,
      "minimum": 1,
      "maximum": 100
    },
    "includeAdversarial": {
      "type": "boolean",
      "description": "Includes prompt-injection and noisy data submitter personas in the cohort.",
      "default": true
    },
    "locale": {
      "type": "string",
      "description": "Target language code for synthetic persona names and prompt payloads (en, es, de, fr, pt, ja, zh; supports BCP-47 tags like pt-BR or zh-CN). Defaults to 'en'.",
      "pattern": "^([eE][nN]|[eE][sS]|[dD][eE]|[fF][rR]|[pP][tT]|[jJ][aA]|[zZ][hH])([-_][a-zA-Z0-9]{2,8}){0,3}$",
      "default": "en"
    }
  },
  "required": [
    "domain"
  ]
}

Output Schema

{
  "type": "object",
  "properties": {
    "domain": {
      "type": "string"
    },
    "locale": {
      "type": "string"
    },
    "totalPersonas": {
      "type": "integer"
    },
    "personaTypes": {
      "type": "object"
    },
    "ipipNeoCoverage": {
      "type": "string"
    },
    "cohort": {
      "type": "array"
    }
  },
  "required": [
    "domain",
    "totalPersonas",
    "personaTypes",
    "ipipNeoCoverage",
    "cohort"
  ]
}
🟢systemores_inject_environment_fault(faultType, targetToolName, failureProbability)

Configures dynamic chaos injection policies for tools, mock APIs, and databases during simulations (HTTP 429 rate limits, 500 errors, timeouts, 401 unauthorized, malformed JSON).

Input Schema

{
  "type": "object",
  "properties": {
    "faultType": {
      "type": "string",
      "description": "The specific chaos fault to inject.",
      "enum": [
        "rate_limit_429",
        "server_error_500",
        "timeout",
        "unauthorized_401",
        "malformed_json"
      ]
    },
    "targetToolName": {
      "type": "string",
      "description": "The target tool name to attach fault injection policy to (or 'all_tools')."
    },
    "failureProbability": {
      "type": "number",
      "description": "Probability of fault triggering per invocation (0.01 to 1.0).",
      "default": 0.5,
      "minimum": 0.01,
      "maximum": 1
    }
  },
  "required": [
    "faultType",
    "targetToolName"
  ]
}

Output Schema

{
  "type": "object",
  "properties": {
    "policyId": {
      "type": "string"
    },
    "targetToolName": {
      "type": "string"
    },
    "faultType": {
      "type": "string"
    },
    "failureProbability": {
      "type": "number"
    },
    "simulatedLatencyMs": {
      "type": "integer"
    },
    "mockResponseBody": {
      "type": "string"
    },
    "activeDurationSeconds": {
      "type": "integer"
    },
    "recommendation": {
      "type": "string"
    }
  },
  "required": [
    "policyId",
    "targetToolName",
    "faultType",
    "failureProbability",
    "simulatedLatencyMs",
    "mockResponseBody",
    "activeDurationSeconds",
    "recommendation"
  ]
}
🟢systemores_analyze_trace_telemetry(otelTraceJson, maxCostThresholdUsd, latencyThresholdMs)

Ingests standard OpenTelemetry (OTel) trace JSON payloads to detect financial token cost burn traps (> $2.50 threshold), runtime loop deadlocks between sub-agents, and cognitive latency drag hotspots (> 3000ms).

Input Schema

{
  "type": "object",
  "properties": {
    "otelTraceJson": {
      "type": "string",
      "description": "JSON string containing OpenTelemetry trace spans or MCP telemetry log data."
    },
    "maxCostThresholdUsd": {
      "type": "number",
      "description": "Maximum allowable cost in USD per session before triggering burn trap alert.",
      "default": 2.5
    },
    "latencyThresholdMs": {
      "type": "integer",
      "description": "Latency hotspot threshold in milliseconds.",
      "default": 3000
    }
  },
  "required": [
    "otelTraceJson"
  ]
}

Output Schema

{
  "type": "object",
  "properties": {
    "traceId": {
      "type": "string"
    },
    "traceContextPropagated": {
      "type": "boolean"
    },
    "totalSpansEvaluated": {
      "type": "integer"
    },
    "estimatedCostUsd": {
      "type": "number"
    },
    "costBurnTrapDetected": {
      "type": "boolean"
    },
    "loopDeadlocksDetected": {
      "type": "array"
    },
    "latencyHotspotsMs": {
      "type": "array"
    },
    "rating": {
      "type": "string",
      "enum": [
        "OPTIMIZED",
        "WARNING",
        "CRITICAL"
      ]
    },
    "recommendations": {
      "type": "array"
    }
  },
  "required": [
    "traceId",
    "traceContextPropagated",
    "totalSpansEvaluated",
    "estimatedCostUsd",
    "costBurnTrapDetected",
    "loopDeadlocksDetected",
    "latencyHotspotsMs",
    "rating",
    "recommendations"
  ]
}
⚪systemores_export_dossier(auditSessionId, targetFrameworks, companyName, locale, systemPrompt, ...)

Aggregates findings from graph audits, synthetic cohort simulations, chaos tests, and trace analytics into signed zero-retention compliance dossiers with verifiable certificate links and live SVG trust badges for EU AI Act (EU 2024/1689), NIST AI RMF 1.0, EEOC/ECOA, and SOC 2 AI Addendums.

Input Schema

{
  "type": "object",
  "properties": {
    "auditSessionId": {
      "type": "string",
      "description": "Unique identifier for the completed audit session."
    },
    "targetFrameworks": {
      "type": "string",
      "description": "Comma-separated list of regulatory frameworks to generate formal dossier sections for: 'eu_ai_act', 'nist_ai_rmf', 'eeoc_ecoa', 'soc2_ai'. Defaults to all if omitted."
    },
    "companyName": {
      "type": "string",
      "description": "Legal entity or client name for dossier attribution.",
      "default": "Enterprise Client"
    },
    "locale": {
      "type": "string",
      "description": "Target language code for human-readable dossier prose and executive summary (en, es, de, fr, pt, ja, zh; supports BCP-47 tags like pt-BR or zh-CN; canonical JSON fields and SHA-256 hash remain invariant). Defaults to 'en'.",
      "pattern": "^([eE][nN]|[eE][sS]|[dD][eE]|[fF][rR]|[pP][tT]|[jJ][aA]|[zZ][hH])([-_][a-zA-Z0-9]{2,8}){0,3}$",
      "default": "en"
    },
    "systemPrompt": {
      "type": "string",
      "description": "Optional agent system prompt evaluated strictly in memory (never stored) to compute differentiated per-framework scores via Murray Needs-Press, drift, bias, and tool-discipline diagnostics. Maximum 8,000 characters."
    },
    "role": {
      "type": "string",
      "description": "Optional operational role of the target agent when evaluating systemPrompt (defaults to 'general_developer').",
      "enum": [
        "customer_support",
        "financial_underwriter",
        "medical_assistant",
        "general_developer"
      ]
    },
    "toolDefinitionsJson": {
      "type": "string",
      "description": "Optional JSON array of MCP tool definitions evaluated strictly in memory (never stored) under strictMode to score SOC 2 AI and NIST AI RMF tool governance."
    },
    "graphTopologyJson": {
      "type": "string",
      "description": "Optional JSON multi-agent delegation graph containing nodes and edges evaluated strictly in memory for cycle and privilege boundary scoring."
    },
    "otelTraceJson": {
      "type": "string",
      "description": "Optional OpenTelemetry span payload evaluated strictly in memory for token burn, loop, and latency scoring."
    },
    "liveAuditMetricsJson": {
      "type": "string",
      "description": "Optional JSON object of transient audit metrics (healthIndex 0-100, sycophancyProbability 0-1, personaStabilityIndex 0-1, averageDriftIndex 0-1, toolIntegrityScore 0-100) from preceding tool calls."
    }
  },
  "required": [
    "auditSessionId"
  ]
}

Output Schema

{
  "type": "object",
  "properties": {
    "dossierId": {
      "type": "string"
    },
    "timestamp": {
      "type": "string"
    },
    "companyName": {
      "type": "string"
    },
    "auditSessionId": {
      "type": "string"
    },
    "locale": {
      "type": "string"
    },
    "frameworksEvaluated": {
      "type": "array"
    },
    "certificationStatus": {
      "type": "string",
      "enum": [
        "CERTIFIED",
        "CONDITIONAL_APPROVAL",
        "NON_COMPLIANT"
      ]
    },
    "executiveSummary": {
      "type": "string"
    },
    "frameworkScores": {
      "type": "object"
    },
    "overallScore": {
      "type": "integer"
    },
    "scoringMode": {
      "type": "string",
      "enum": [
        "dynamic_evidence",
        "baseline_attestation"
      ]
    },
    "engineVersion": {
      "type": "string"
    },
    "zeroRetentionVerified": {
      "type": "boolean"
    },
    "sha256IntegrityHash": {
      "type": "string"
    },
    "verificationUrl": {
      "type": "string"
    },
    "badgeSvgUrl": {
      "type": "string"
    },
    "badgeMarkdown": {
      "type": "string"
    },
    "dossierMarkdown": {
      "type": "string"
    }
  },
  "required": [
    "dossierId",
    "timestamp",
    "companyName",
    "auditSessionId",
    "frameworksEvaluated",
    "certificationStatus",
    "executiveSummary",
    "frameworkScores",
    "overallScore",
    "scoringMode",
    "engineVersion",
    "zeroRetentionVerified",
    "sha256IntegrityHash",
    "verificationUrl",
    "badgeSvgUrl",
    "badgeMarkdown",
    "dossierMarkdown"
  ]
}
🟢systemores_audit_turn_live(systemPrompt, userMessage, agentResponse, proposedToolCall)

Performs real-time (~250ms) operational and behavioral turn auditing using Systemores System One. Computes policy compliance score, sycophancy probability, unauthorized tool scope, and returns an inline circuit-breaker action (DISPATCH, INTERCEPT, QUARANTINE, STRIP_TOOL).

Input Schema

{
  "type": "object",
  "properties": {
    "systemPrompt": {
      "type": "string",
      "description": "The behavioral blueprint, constraints, or security policies of the agent under audit."
    },
    "userMessage": {
      "type": "string",
      "description": "The prompt or message sent by the user in this interaction turn."
    },
    "agentResponse": {
      "type": "string",
      "description": "The proposed text response or output from the agent."
    },
    "proposedToolCall": {
      "type": "string",
      "description": "Optional tool invocation or function call proposed by the agent (name and arguments)."
    }
  },
  "required": [
    "systemPrompt",
    "userMessage"
  ]
}

Output Schema

{
  "type": "object",
  "properties": {
    "engine": {
      "type": "string"
    },
    "status": {
      "type": "string",
      "enum": [
        "AUDITED",
        "FALLBACK"
      ]
    },
    "latencyMs": {
      "type": "number"
    },
    "tokens": {
      "type": "object"
    },
    "metrics": {
      "type": "object"
    },
    "verdict": {
      "type": "string",
      "enum": [
        "PASSED",
        "DRIFT_WARNING",
        "CRITICAL_BREACH",
        "TOOL_SCOPE_VIOLATION"
      ]
    },
    "circuitBreakerAction": {
      "type": "string",
      "enum": [
        "DISPATCH",
        "INTERCEPT",
        "QUARANTINE",
        "STRIP_TOOL"
      ]
    },
    "autoRecoveryRecommendation": {
      "type": "string"
    }
  },
  "required": [
    "engine",
    "status",
    "latencyMs",
    "metrics",
    "verdict",
    "circuitBreakerAction"
  ]
}

Community

Rate this Server

Evidence

Recent observations

verifiedversion not recorded12 tools