Developer Experience

Put your docs inside the IDE via MCP

Cursor, Claude, and other MCP clients answer from your docs.

The pain

Developers now ask their AI coding assistant instead of reading your docs — and the assistant answers from stale training data, hallucinating deprecated endpoints and wrong parameter names. You get support tickets for APIs you fixed a year ago.

How Kelu solves it

Kelu exposes each knowledge base as an MCP server with search_docs, ask_question, and list_sources tools. Developers add one endpoint to Cursor, Claude, or any MCP client, and their AI assistant retrieves your current documentation instead of guessing — grounded answers, right in the editor.

How it works

1Each Kelu knowledge base exposes an MCP server endpoint out of the box.
2Developers add the endpoint to Cursor, Claude, or any MCP client, authenticated with a client key.
3The AI assistant calls search_docs, ask_question, and list_sources against your current index.
4Answers in the editor are grounded in your live docs instead of stale training data.

What you get

Stop AI coding assistants from hallucinating deprecated endpoints and wrong parameters.

Meet developers where they already ask questions — inside the IDE.

Keep in-editor answers current with your docs via the always-synced index.

Ship MCP support with zero extra infrastructure on your side.

Features that make it work

MCP server per knowledge base

search_docs / ask_question tools

Client-key auth

Always-current index

Frequently asked questions

What exactly does the MCP server expose?

A JSON-RPC MCP endpoint per knowledge base with three tools: search_docs (hybrid search over your index), ask_question (full RAG answer with citations), and list_sources. Any MCP-aware client can use them.

How do developers authenticate?

With a client key issued per knowledge base. Keys can be scoped, so the MCP surface only reaches the knowledge you intend to expose.

Which tools can connect to it?

Any MCP client — Cursor, Claude, and other MCP-aware editors and assistants, plus your own tooling.

How long does setup take?

Most teams are live the same day. You connect a source — a docs URL, GitHub repo, Notion workspace, or a file upload — and Kelu crawls, normalizes, chunks, and embeds the content automatically. The website widget is a single script tag, and the Slack/Discord bots install in a few clicks. There is no model to fine-tune and no infrastructure to run.

Are answers backed by citations?

Yes. Every answer includes a structured citations list with the URL, title, and excerpt of each source used, and the answer text cites sources inline. Users can click through to verify any claim against the original document — which is what makes the answers trustworthy enough for support, sales, and internal use.

Is my content used to train AI models?

No. Your indexed content, embeddings, and conversations are isolated to your workspace and are only used to answer your own questions. Content is never shared across customers, and data is encrypted at rest and in transit. Enterprise plans add OIDC/SAML SSO, audit logs, and data-residency options.

Which knowledge sources can I connect?

25 source types: websites, GitHub (repos, issues, pull requests, discussions), Notion, Confluence, Google Drive, OneDrive, Zendesk, Jira, Jira Service Management, Linear, Salesforce (objects and cases), Slack, Discord, Discourse, Stack Overflow, YouTube, OpenAPI specs, S3, helpdesk tickets, curated Q&A, and direct file uploads (PDF, DOCX, XLSX, CSV, Markdown, TXT, HTML). Sources sync on a schedule, and GitHub sources can re-index automatically via webhook on every push.

What happens when the AI is not confident in an answer?

Answers are confidence-scored. When retrieval confidence falls below your threshold, Kelu returns a transparent, configurable fallback — typically pointing users to your support channel — instead of guessing. Those unanswered questions are logged in knowledge gap analytics so you can close the gap in your docs.