NeetoKB MCP and CLI: let AI tools help write and run your knowledge base

What is NeetoKB MCP?

NeetoKB MCP lets AI tools like Claude and ChatGPT work with your NeetoKB knowledge base. Once it's connected, you can ask your AI assistant what your help center already covers, draft an article from a support conversation, clean up outdated content, and publish, in the same conversation where you're already working.

MCP stands for Model Context Protocol, an open standard for connecting AI tools to the apps you use. You don't need to know anything about it. You connect once, sign in to NeetoKB, and start asking.

Your AI tool talks to the NeetoKB MCP server. NeetoKB hosts it, so there's nothing to install or keep updated, and every workspace uses the same address:

https://connect.neetokb.com/mcp/messages

When you connect, you sign in with your NeetoKB account and pick your workspace. The assistant then works as you: it sees only the articles and categories you can see, and does only what your role allows. It can search and read articles, create and update them, share them with private links, and see your categories, authors, and team members. You never name a tool; you ask in plain language. The full list is in the NeetoKB MCP tools reference.

NeetoKB also has a command-line tool, the NeetoKB CLI. It works with the same articles, but runs in your terminal, in scripts, in CI, and in scheduled jobs. It's open source, so you can read exactly what it does, change it, and build on it.

Real prompts for real work

These are real prompts for real knowledge base work. Copy any of them and try it with Claude, ChatGPT, or another AI tool. For repeatable jobs in automation, we have included CLI examples too.

Ask NeetoKB questions

NeetoKB isn't only something your AI assistant can write to. It's something you can ask questions about. Try these:

Do we have an article that explains how to reset a password?

What are our ten most-viewed articles?

Which articles are still drafts, and who wrote them?

Which categories don't have a single published article?

Which articles haven't been updated in more than a year?

Which articles still mention "Acme Classic"?

Who has written the most articles?

Which articles are in private categories?

Which articles do we recommend on the billing page?

The assistant reads your articles and categories and works out the answer. For questions that span thousands of articles, the CLI is the better fit. See MCP or CLI.

Find gaps and fix content

Find what's missing

Paste this week's most common support questions and ask:

Which of these questions don't have a knowledge base article yet? For each one that's missing, tell me which category it belongs in.

Instead of guessing, you get a list of real gaps, ranked by how often customers ask.

Check before answering a customer

A customer asks whether we support single sign-on with Okta. What does our knowledge base say? Quote the relevant part and give me the link.

Find outdated content

Find published articles that still mention the old pricing plans. List each one with the sentence that needs changing. Don't change anything yet.

Fix a product name everywhere

We renamed "Acme Classic" to "Acme Standard". Find every article that uses the old name, show me the changes you'd make, and update them once I approve.

Write and publish

Turn a support conversation into an article

A customer and your team just worked out a tricky setup in a ticket. Paste the conversation and ask:

Write a help article from this conversation, for other customers with the same problem. Put it under Integrations, then Slack, and leave it as a draft.

The assistant writes the article, files it in the right category, and creates the category if it doesn't exist yet. Nothing is published until you say so.

Turn release notes into help content

Here are the release notes for version 4.2. Which existing articles need updating because of these changes? Draft a new article for the Custom domains feature.

Rewrite for clarity

Rewrite the "Getting started" article in simpler language, with numbered steps. Show me the new version before you save it.

Review and publish drafts

Go through our draft articles. Tell me which ones look ready to publish and which need more work, and why. Publish the ready ones after I confirm.

Share an article privately

Give me a private link to the Enterprise SSO setup article that expires in seven days, so I can send it to a prospect.

A new private link replaces any existing one, so the assistant will tell you if an older link will stop working.

Manage your team

Invite writers

Invite oliver@example.com and sam@example.com to NeetoKB as Editors.

The role you name must already exist in your workspace.

Remove someone who left

Sam has left the company. Deactivate sam@example.com in NeetoKB.

Their articles stay published and stay credited to them.

Automate with the CLI

Why have a CLI when MCP exists? Because some jobs shouldn't need an AI at all. These run on a schedule or in CI, with nothing but the neetokb binary and the tools you already use, such as jq, curl, and your cloud CLI. No model account, no cost per run, and the same result every time.

Publish release notes from CI

When a release ships, publish its notes as an article straight from your pipeline.

neetokb articles create \
  --title "What's new in version $VERSION" \
  --category "Release notes" \
  --html-content "$(pandoc CHANGELOG.md)" \
  --state published

Back up every article nightly

page=1
while :; do
  out=$(neetokb articles list --page "$page" --page-size 100 --json)
  echo "$out" | jq -r '.data[].id' | while read -r id; do
    neetokb articles show "$id" --json | jq -c '.data.article'
  done
  [ "$(echo "$out" | jq '.pagination.total_pages')" -le "$page" ] && break
  page=$((page + 1))
done > articles.jsonl

aws s3 cp articles.jsonl "s3://your-backup-bucket/neetokb/$(date +%F).jsonl"

Report the most-viewed articles

Every Monday, post the ten most-viewed published articles to your team channel. The first step saves every published article to a file, and the next job reuses it.

page=1
while :; do
  out=$(neetokb articles list --state published --page "$page" --page-size 100 --json)
  echo "$out" | jq -c '.data[]'
  [ "$(echo "$out" | jq '.pagination.total_pages')" -le "$page" ] && break
  page=$((page + 1))
done > published.jsonl

jq -s '{text: ([sort_by(-.unique_views_count)[:10][] | "\(.unique_views_count)  \(.title)"] | join("\n"))}' published.jsonl \
  | curl -s -X POST -H "Content-Type: application/json" -d @- "$SLACK_WEBHOOK_URL"

Flag articles nobody has touched in a year

cutoff=$(date -d "1 year ago" +%F)   # on macOS: date -v-1y +%F

jq -r --arg cutoff "$cutoff" 'select(.updated_at < $cutoff) | "\(.updated_at[:10])  \(.title)"' published.jsonl

Check that a search still finds the right answer

Run this in CI after content changes, and fail the build if a key question stops finding its article.

neetokb search --search-term "reset password" --json \
  | jq -e '.data[] | select(.title == "How to reset your password")' > /dev/null \
  || { echo "Search for 'reset password' no longer finds the article"; exit 1; }

Refresh a private link every week

neetokb articles unlisted-links regenerate enterprise-sso-setup --expiration-type seven_days

Add new writers

neetokb team-members create --emails oliver@example.com,sam@example.com --role Editor

Let a coding agent build it for you

You don't have to write these scripts yourself. Claude Code, Codex, or another coding agent can write the automation and use the open-source NeetoKB CLI to work with your knowledge base. Run neetokb setup claude (or setup codex, setup cursor) once, then ask:

Our docs live as Markdown files in the docs folder. Write a script that uses the NeetoKB CLI to publish any new or changed file as an article, and run it in our CI on every merge to main.

Using the NeetoKB CLI, write a weekly job that lists draft articles older than 30 days and posts them to our Slack channel.

You get the convenience of a chat, and the result is a script you can read, run, and commit.

MCP or CLI: which should you use?

Both work with the same articles and categories. Pick based on how the work reaches NeetoKB.

Use MCP when:

• The source material is in your chat. A support thread, release notes, or a pasted document becomes an article with no copying.
• The work needs reading and writing. Finding gaps, rewriting for clarity, and reviewing drafts need someone to read the content.
• You're asking a question. "Do we have an article about Okta?" needs someone to search and explain.
• You don't use a terminal. NeetoKB hosts the server. There's nothing to install.

Use the CLI when:

• No AI should be in the loop. A CI step or cron job runs the CLI with nothing but the binary. No model account, no cost per run.
• Your docs live in a repository. Publish from your build pipeline, the same way every time.
• The output feeds another program. The CLI prints raw JSON for jq, a report, or your own script.
• The run has to be repeatable. A command records exactly what ran and does the same thing every time.

You can also have both. With neetokb setup claude, your AI assistant drives the CLI for you, so a plain-language request still ends in a command you can read and reuse.

Connect NeetoKB to your AI tool

Claude

1. Open NeetoKB in the Claude directory and click Connect to Claude.
2. Sign in to NeetoKB and choose your workspace and permissions.

This works in Claude on the web, desktop, and mobile. On a Claude Team or Enterprise plan, an admin may need to turn on NeetoKB under Organization settings, then Connectors.

ChatGPT

1. In ChatGPT, open Settings, then Security and login, and turn on Developer mode.
2. Go to chatgpt.com/plugins and click +.
3. Name it NeetoKB, set the connection to Server URL, and paste https://connect.neetokb.com/mcp/messages.
4. Leave authentication on OAuth, accept the notice, and click Create. Then sign in to NeetoKB.

Claude Code

claude mcp add --transport http neetokb https://connect.neetokb.com/mcp/messages

Then start Claude Code, run /mcp, pick neetokb, and sign in.

Cursor, Codex, VS Code, Windsurf, and Gemini CLI

Each of these connects to the same server address. The NeetoKB MCP setup guide has the exact configuration for each one.

To check the connection, ask your assistant: "Search my NeetoKB articles for billing."

Install the NeetoKB CLI

macOS with Homebrew:

brew install neetozone/tap/neetokb

macOS or Linux:

curl -fsSL https://neetokb.com/cli/install.sh | sh

Windows (PowerShell):

irm https://neetokb.com/cli/install.ps1 | iex

Then sign in. This opens your browser, and there's no API key to manage:

neetokb login --subdomain your-workspace

To teach your AI coding assistant how to use the CLI, run one of these from your project folder:

neetokb setup claude      # Claude Code
neetokb setup cursor      # Cursor
neetokb setup codex       # Codex
neetokb setup copilot     # GitHub Copilot
neetokb setup gemini      # Gemini CLI
neetokb setup windsurf    # Windsurf

Every command supports --json for scripts and --toon, a compact format that uses fewer tokens when an AI tool reads the output. See the NeetoKB CLI reference for every command.

Permissions and safety

• Your assistant works as you. When you sign in with NeetoKB, it sees only the articles and categories you can see, and private categories stay hidden from anyone who can't open them.
• Publishing follows your roles. Creating and editing drafts needs permission to modify articles, and publishing needs permission to manage them.
• You choose the access. Read is always on. Create and update lets it write, update, and publish articles, and Delete only lets it deactivate team members. Leave both off for a read-only connection.
• Articles can't be deleted through AI. The only removal is deactivating a team member, which keeps their articles and credit.
• You can disconnect anytime. Remove the NeetoKB connector from your AI tool and its access ends.

New articles start as drafts unless you ask for them to be published. A good habit is to ask your assistant to show you its changes before it updates or publishes anything, as several prompts above do.

Open source

The NeetoKB CLI is open source on GitHub. You can read every command, see exactly what it sends to NeetoKB, and adapt it to your own workflows. It's written in Go and runs on macOS, Linux, and Windows.

The CLI is built on neeto-cli-commons, the shared foundation behind every Neeto CLI, so what you learn in one works the same way in the others.

Found a bug or want a new command? Open an issue or a pull request.

Frequently asked questions

Do I need to know what MCP is?

No. Connect NeetoKB to your AI tool once, sign in, and ask in plain language.

Can ChatGPT write articles in NeetoKB?

Yes. Connect ChatGPT to the NeetoKB MCP server and it can search your articles, draft new ones, update existing ones, and publish them, within your permissions. Today you add NeetoKB to ChatGPT as a custom connector in Developer mode. See Connect NeetoKB to your AI tool.

Can Claude search my NeetoKB knowledge base?

Yes. Connect NeetoKB from the Claude directory and you can ask Claude what your help center says about a topic, get the article link, and draft or update articles in plain language.

Will AI publish articles without asking?

Only if you ask it to. New articles start as drafts by default, and publishing needs a role that's allowed to publish. You can also connect with read-only access.

Can I use NeetoKB with Claude Code?

Yes, in two ways. Add the NeetoKB MCP server with claude mcp add, or run neetokb setup claude so Claude Code can drive the NeetoKB CLI. You can use both.

Can I use NeetoKB MCP with Cursor or Codex?

Yes. Cursor, Codex, VS Code with GitHub Copilot, Windsurf, and Gemini CLI all connect to the same NeetoKB MCP server. The setup guide has the configuration for each.

Does NeetoKB MCP require an API key?

No. You sign in with your NeetoKB account and approve access, so there's no key to copy or rotate. Developer tools such as Claude Code, Cursor, and Codex can use a NeetoKB API key instead if you prefer, but a key reaches the whole workspace rather than acting as you.

Can I give AI read-only access to NeetoKB?

Yes. Read access is always on, while Create and update, and Delete, are separate permissions. Leave them unchecked if you want your AI assistant to search and read articles without changing anything.

Can I publish NeetoKB articles from my docs repository?

Yes. The NeetoKB CLI creates and updates articles from your terminal or CI, so you can publish Markdown from a repository on every merge.

Can I automate NeetoKB without using AI?

Yes. The NeetoKB CLI works directly from your terminal, shell scripts, cron jobs, and CI workflows. You don't need Claude, ChatGPT, or another AI service to use it.

Is the NeetoKB CLI open source?

Yes. It's on GitHub, and you can use it, read it, and change it.

Where is the NeetoKB MCP server hosted?

NeetoKB hosts the MCP server at https://connect.neetokb.com/mcp/messages. You don't need to download or run a server yourself.

Does NeetoKB MCP work on the free plan?

Yes. The MCP server and the CLI are included in every NeetoKB plan, including the free plan.