llmsecurity-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
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
| Capability | Details |
|---|---|
| tools | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| scan_agent_configA | Scan AGENTS.md, CLAUDE.md, SKILL.md, .claude, and .codex files locally for prompt injection. |
| check_tool_call_safetyC | Classify a proposed tool call as allow, ask, or deny before execution. |
| scan_prompt_for_injectionC | Run existing PromptGuard patterns against supplied instructions. |
| scan_rag_sourceA | Scan local RAG source documents for prompt injection, secrets, and PII before indexing. |
| map_security_coverageB | Map a LLMSecurity scan report to OWASP LLM 2025, NIST AI RMF, and MITRE ATLAS coverage evidence. This is not a compliance certification. |
| scan_skill_trustC | Build a local trust inventory for skills, MCP package manifests, and risky install scripts. |
| analyze_model_abuseB | Analyze customer-controlled model API telemetry for burst, cost-drain, repeated-query, and auth-failure signals. Never supply prompt bodies. |
| check_mcp_policyC | Evaluate an MCP tool call against a local allowlist/approval/destination policy. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 8 tools
Most tools have clear targets (config files, prompts, RAG docs, skills, telemetry), but check_tool_call_safety and check_mcp_policy both evaluate tool calls against policy, so an agent could easily pick the wrong one.
All tools follow snake_case verb_object naming (scan_*, check_*, map_*, analyze_*) with no mixed conventions, making the set predictable and easy to navigate.
Eight tools is well within the ideal range, and each tool covers a distinct security workflow without feeling bloated or too thin.
The set covers scanning, policy enforcement, trust inventory, and abuse detection, but lacks a remediation/response tool and some scans only apply to pre-indexing or pre-execution stages.