"How to allow an LLM to ask questions to a user" matching MCP servers:

Enables AI agents to request an independent second opinion from a different frontier model, with PII scrubbing before transit, live web search grounding, caching, and multi-provider failover.

An MCP server that sends plain-text Gmail messages only to a predefined allowlist, using generated recipient IDs instead of email addresses to enforce security.

Exposes user, membership, and role administration tools via MCP, acting as a thin passthrough to a coordinator API with RBAC enforced by the coordinator using caller bearer tokens.

MCP server enabling Claude Code to capture screenshots of Windows applications through WSL2, allowing AI to diagnose UI errors and verify layouts.

MCP server demonstrating Streamable HTTP with Tools, Resources, and Prompts for user registration, login, and profile management via a REST API.

Enables cloud-based MCP clients to delegate privacy-sensitive or large-output tasks to a local LLM, returning sanitized placeholders or digested summaries so private data never leaves the user's network.

MCP server toolkit with payload library, WAF bypass generator, rate limiter, plugin system, and inter-MCP bridge for security testing.

Enables LLMs to propose UPDATE/DELETE SQL that is run in a transaction, measured, and rolled back, requiring human approval before applying to prevent unauthorized changes.

Provides a governance gateway for LLM and agent traffic via MCP, blocking prompt injection, redacting PII, enforcing per-tenant token budgets, and caching repeat questions semantically. Enables AI agents to route model calls through the same guardrails instead of calling providers directly.

This MCP server provides secure access to databases for AI agents, enforcing authentication, authorization, human approval, logging, and notifications to prevent dangerous actions.

Enables ZenTao user administration, including token retrieval, user queries, and user creation/update/delete, over stdio or Streamable HTTP transports.

Enables a cloud orchestrator to query a local Ollama model over Tailscale or a LAN as a side-channel for drafts, summaries, and other helper output, with tools to list available models and to redact text. Answers leaving the server are passed through output-only DLP that masks card numbers, IBANs, emails, phone numbers, API keys, and similar secrets before they reach the cloud.

Exposes an OpenAI- and Anthropic-compatible HTTP, SSE, and stdio gateway that wraps multiple subscription CLIs, adding prompt-injection defense, PII redaction, cost-aware routing, and reasoning-trace capture for MCP-compatible clients like Claude Desktop and Cursor.

Tells you how you and your Claude agent actually work together. Reads CLAUDE.md, hooks, skills, memory + scheduled tasks and writes you a letter. Local-only, no API keys, no data leaves your machine.

Exposes LLM output validation (schema compliance and prompt-injection detection) as an MCP tool, enabling any MCP-compatible agent to apply real-time guardrails without importing the codebase.

OpenAI-compatible proxy with PII scrubbing, audit logging, and model routing. Drop-in replacement — set OPENAI_BASE_URL to point to this server. Supports GPT-4o, Claude, Gemini, and Llama via single endpoint.

Provides a validated MCP server with strict schema-checked tools, alongside security gateway filtering, streaming PII redaction, and rate-limited model fallback routing.

Authenticates users via Google OAuth and returns the authenticated user's display name through the get_current_user tool. Supports deployment on Node.js and Cloudflare Workers.

Enables AI agents to access only task-relevant data with secrets hidden and personal details pseudonymized, while enforcing destination policies to prevent data leakage even under prompt injection.

Experimental read-only MCP for Supabase application data: memory_search, memory_get and memory_list_recent under user-bound PostgreSQL RLS. Setup and demo: https://github.com/jryski/Supabase_user_MCP/blob/main/docs/GETTING_STARTED.md . Public preview is POSIX-only and synthetic-only; native Windows, hosted OAuth, one-click installation and writes are not available.