- Conduid
- Marketplace
- #llm-security
MCP servers tagged llm-security
557 live MCP servers tagged "llm-security", ranked by trust score. Tags come from package metadata and repository topics, so this list covers servers across every category that work with llm-security.
Tulip Agents
The agent framework where the model never holds the trigger — every consequential action clears your policy first, waits for a human when it matters, and lands on a record you can…
Mcpguard Go
mcpguard-go: local AI agent security scanner (Go, zero-dep single binary)
Poison Scan
CLI scanner for MCP server tool-poisoning & rug-pull signatures (hidden instructions, invisible-Unicode, multilingual/Turkish-casefold evasions). SARIF + rug-pull diff. Extends mc…
SentinelRAG
A policy-aware hybrid RAG agent for secure question answering over code repositories and documentation, with BM25/FAISS retrieval, MCP-style tool routing, and multi-model evaluati…
Promptmirror
Prompt-injection & indirect-injection scanner for any LLM context input
Nightprism
MCP-native security scanner that asks "should I ship tonight?" for AI-generated code.
Sentinel Trace
🛡️ Catch unsafe AI agent actions before they execute — explainable risk scoring, injection detection, and real-time containment powered by SigNoz & OpenTelemetry.
Toolproof
Trust grades, capability lockfiles, and signed evidence for the AI tool supply chain. Scan any MCP server or API before your agent connects. Paid by nobody we scan.
Jak Shield
Secure AI agent tool calls by blocking destructive actions, redacting sensitive data, and enforcing human oversight with this universal security gateway.
Soc Triage Agent
Local defensive AI agent that triages synthetic SOC identity alerts (LangGraph + MCP), with a governance-as-code layer - security manifest, control matrix (OWASP LLM Top 10 / MITR…
AI Integration Health Check Demo
Demo repo for an AI/API integration health-check freelance service: IDOR/SQLi/secret-logging fixes side by side (broken vs main), zero-dependency idempotency example, audit report…
Bounded Local AI Demo
OS-enforced isolation demo for bounded local AI: an agent reaches a private vault only through the severino-vault-mcp sensitivity gate, with no filesystem bypass.
SkillWatch
Independent, local-only watcher for the AI-skill supply chain: flags bait-and-switch changes in the pages skills and MCP tools reference. Sends nothing anywhere. Python CLI.
Policies
78 Cedar policies and 369 rules governing AI coding agents. Every rule traces to a real incident, published CVE, or compliance framework requirement (SOC 2, NIST, ISO 27001, E…
Agent Tool Audit
Audit AI agent tool manifests for least-privilege, schema, and capability-combination risks.
Trifecta Guard
Runtime security middleware that blocks lethal prompt-injection capability chains in AI agents.
Scrim
Claude Code plugin that keeps secrets and PII out of the model's context window by reversibly tokenizing them at the tool boundary — the agent works with masked content, real valu…
Security Docker Gateway
Vulnerable MCP server + layered defenses (static scan, runtime interceptor, Docker MCP Gateway).
Pnx MCP Audit
Security audit CLI for Model Context Protocol (MCP) servers. Zero deps, offline, cross-platform.
Bert Bouncer
On-device prompt-injection firewall for Claude Code. Hybrid DeBERTa + masked-LM detection with span localization, content sanitization, and a reproducible red-team evaluation. No…
AI Agent Security Control Plane
Defense-in-depth security control plane for autonomous AI agents using MCP, OPA/Rego, human approval, adaptive risk, detection, and automated containment.
Two Stage Sanitization Architecture
Two-stage security architecture to mitigate indirect prompt injection attacks in rich content via privilege separation
Opa Sidecar A2a
Reference implementation of chain-aware authorization for agent-to-agent (A2A) traffic, using Open Policy Agent.
Model Trust Gate
A fail-fast, layered decision gate for deciding whether to trust an AI model for a specific use. A methodology that composes existing standards (NIST, ISO 42001, CSA AICM, OWASP,…
Vetpkg
Stop AI coding agents installing packages that do not exist. MCP server + CLI for detecting hallucinated and slopsquatted npm packages. Zero dependencies.
Objauthz Lab
Vulnerable-by-design MCP server for learning object-level / cross-tenant authorization (BOLA/IDOR) bugs, plus Semgrep rules that find the same shapes in JavaScript/TypeScript, Pyt…
Agenthive
Multi-agent attack simulation framework for AI systems. 8 attack categories, simulation engine, CLI, SARIF/JSON/HTML reporting, mcp-taxonomy integration.
Damn Vulnerable Agent Asset Corpus
Damn Vulnerable Agent Asset Corpus: benchmark fixtures for agent-asset scanners and AAC release assurance.
Dvarapala
द्वारपाल — gatekeeper between your LLM and third-party MCP servers. Reads every JSON-RPC message, denies what fails policy, redacts what leaks, audits the rest.
Adversa
LLM red-team harness — OWASP LLM Top 10 + MITRE ATLAS attack packs
Hades Blackwatch
Hades Blackwatch is an advanced AI security and code-review skill for auditing LLM, agentic, MCP, and full-stack applications using OWASP LLM/Agentic Top 10, AIVSS, MITRE ATLAS, N…
Agentic Compliance Platform
Engineering case study: an agentic AI tier for a denied-party-screening compliance platform — safe, grounded, auditable LLM tooling (RAG, MCP, HITL). Sanitized code + architecture…
Trust Scoreboard
Open-source MCP security dashboard and trust scoreboard for Model Context Protocol servers, AI agents, tools, permissions, identity, provenance and risk.
Zence
Task-scoped context and policy firewall that keeps Claude Code inside the correct client, domain, and environment — using DataHub's metadata graph.
Prompt Injection As An Architectural Primitive
Paper, harness, and raw data: why detection fails against semantic poisoning in agentic LLM pipelines.
Mainsheet
An agents runtime. Define an agent in one file, run it governed, prove what it did. A policy gate decides before every tool call in code the model never sees; budgets cap cost and…
LLM Engineering Portfolio
Thirteen measured experiments on one RAG pipeline: retrieval evaluation, hallucination defense, prompt injection, MCP, and stateful agents. Every number produced by code in this r…
Agentic Redteam Harness
A/B red-team harness for agentic AI: 34 OWASP-LLM / MITRE-ATLAS attacks fired at a multi-agent app with and without a control plane, plus a live malicious-MCP-server demo. Determi…
Whaleguard Redlab
Local-first, scope-guarded AI red-team lab for authorized LLM, Agent and MCP security evaluation.
Trustboost Dev
TrustBoost PII Sanitizer is a pay-per-call "privacy firewall" for autonomous AI agent pipelines, built and operated by an individual developer (Teodoro Crispin, GitHub teodorofodo…
Llmsafe
Static security scanner for Python AI agents, LLM apps, and MCP servers.
Hallumark
LLM hallucination & grounding auditor for RAG systems
Guarded Infra MCP
Read-only Kubernetes and AWS access for AI agents, behind a deny-by-default policy gate with a JSONL audit trail and eval suites that score both per-call block rate and whole-traj…
Splunk MCP Guard
Policy-enforcing proxy in front of any Splunk MCP server
Prompt Shield Security MCP
Deterministic prompt injection classification, secret entropy detection, PII redacting, and synthetic dataset bias auditing.
Security Toolkit
13 atomic MCP tools for AppSec and AI Security: recon, MCP audit, prompt-injection audit, OWASP LLM Top 10
Psp Cdl
Public-domain PSP and CDL standards, reference LLM/MCP proxies, and reproducible workflow-security evaluation. Sponsored by RealflowCloud, Inc.
Cutout
Offensive framework for authorized security testing of agentic AI systems — recon, exploit, persist, and pivot across MCP servers and agent chains.
Guarded SQL MCP
MCP server that gives LLM agents read access to PostgreSQL through a fixed catalog of parameterized, read-only queries. No free-form SQL.
LLM Agent MCP Security Lab
Deliberately vulnerable local lab for hands-on LLM agent & MCP security testing; SSRF, path traversal, SQLi, indirect prompt injection, RAG poisoning, stored XSS and each with a S…