MCP servers · ankit512

Grounded Support Agent

An MCP server for customer-support questions that answers only from knowledge-base passages, with citations, and escalates everything else to a human.

This text was machine-translated from the publisher's own information and has not been checked by a human editor. The publisher's original text is the reference.

This resource was imported automatically from the sources listed below and has not been reviewed by the AI Gate team yet. The description is the publisher's own, as published. Check the repository before installing.

What it does

The agent answers a customer question only when a knowledge-base passage covers it, and it cites that passage with a confidence figure. When the knowledge base only partly covers a question, or does not cover it at all, the agent hands the question to a human and attaches the closest passages or the reason for the handoff. The resolve-or-escalate decision comes from deterministic retrieval and term coverage, checked against explicit thresholds in core/resolver.py, not from a prompt. The model's only role is to word an answer that has already passed that check.

It suits support flows where every answer must trace back to a source, and orchestrators that need a tool which will not invent a resolution. The labelled evaluation set measures honesty under ignorance: out-of-scope, partly out-of-scope, multi-intent and prompt-injection questions must not resolve, and any hallucination fails the run with a non-zero exit code.

The core needs Python 3.9 or later and only the standard library, with no API key and no network access. The knowledge base ships inside the package, so a standalone install needs no repository checkout. The MCP layer is a thin client of the decision engine and computes no answers itself.

Limitations and requirements

Requires Python 3.9 or later. Retrieval uses BM25 and term coverage, so questions worded far from the knowledge base, with heavy paraphrase or synonyms, can score below the threshold and be escalated even though the knowledge base covers them. The publisher states that this failure biases toward escalation. The semantic retriever that could improve recall is documented as a future seam and is not included in this package.

Safety

Our team has not reviewed this resource. Only automated rules checked it; resources with high-risk signals are not published. These are signals, not a guarantee that installing it is safe. Try it in a sandbox project first.

  • Runs code or changes an assistant's behaviour

Sources

  • Official MCP Registry first seen 13 Nov 2026 · last seen 13 Nov 2026 · its page on the source
    io.github.Ankit512/grounded-support-agent

    Governed support MCP: answers only KB-grounded questions with citations, escalates the rest.

    v0.1.0
    As shown on registry.modelcontextprotocol.io when we last read it
MCP servers
awslabs

MCP servers for AWS (AWS Labs)

A large open-source collection of MCP servers from AWS Labs for building and operating on AWS: documentation, infrastructure, databases, AI services, cost and more.

9.9K GitHub starsCloud & DevOpsData & databases
Updated 17 Nov 2026
MCP servers
microsoft

Playwright MCP (browser automation)

Microsoft's MCP server that lets an AI assistant open web pages, click, type and read them through a real browser, using the page structure instead of screenshots.

39.6K GitHub starsTesting & QAWeb, search & browser
Updated 18 Nov 2026
MCP servers
docker

Docker Hub MCP Server

Docker's MCP server that helps an AI assistant find and compare container images on Docker Hub and, with your token, manage your own repositories and tags.

171 GitHub starsCloud & DevOpsSoftware development
Updated 13 Nov 2026
MCP servers
cloudflare

Cloudflare MCP servers

Cloudflare's hosted MCP servers for its products: documentation search, sandbox containers, browser rendering, Radar internet data, logs and analytics, Workers and more.

4.5K GitHub starsCloud & DevOpsSoftware development
Updated 18 Nov 2026