Cactus

ai.heycactusPublished by heycactus.aiOfficial publisherv1.0.0Updated Oct 26, 2026

Search and ask your product's help pages, product map and support conversations in Cactus.

Overview

AI-generated overview

Cactus MCP lets an assistant search a product's help pages, product map, release timeline and past support conversations, and ask cited questions.

What you can do
  • List the team's products and browse their help topics
  • Search and read published help pages
  • Read the product's dated release timeline and the product map built from its code
  • Search and read past customer conversations, and search imported help articles
  • Ask Cactus a question answered from pages, product map, knowledge and conversations, with sources cited
  • Start an investigation into the product's code and knowledge, then read the finding
When to use it
Useful for support, product and docs teams whose product is on Cactus and who want an assistant to answer questions from the help center, product map and past customer conversations instead of guessing. Also handy for checking what changed in recent releases.
Requirements
Remote Streamable HTTP endpoint at mcp.heycactus.ai; no local package. Sign-in is OAuth 2.0 with a Cactus account (Google or GitHub), with dynamic client registration and PKCE. Requires a Cactus team with a product built from a GitHub repository; the product is in closed beta, so access may need a waitlist.
Before you install
Signed-in tools expose internal material: past customer conversations, imported help articles and a product map derived from your codebase, so review who on the team gets access. The ask and start-investigation tools are not read-only, since each runs a new answer or investigation. Public help-center servers need no sign-in, and their contact-team tool sends a message to the team.

Installation

In SourceWeft

  1. Open Cactus in the dashboard and add it to a workspace.
  2. Connect the required credentials. They are stored privately and never shown on public pages.
  3. Enable the server for the chats that should use its tools.

Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.

Use with an agent

Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.

Read https://sourceweft.com/mcp/ai.heycactus%2Fcactus/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.

Read the install guide the agent follows

Other MCP clients

Add this to your client's mcpServers config. The client completes OAuth sign-in on first connect.

{
  "mcpServers": {
    "cactus": {
      "type": "http",
      "url": "https://mcp.heycactus.ai/mcp"
    }
  }
}

README

Cactus MCP server

Cactus (heycactus.ai) learns a product from its codebase, answers its customers live, and keeps its help site and changelog current with every release. Its MCP server lets your team reach that knowledge from Claude, ChatGPT, Cursor, VS Code or any other MCP client.

This repository holds no server code. It lists Cactus's hosted MCP server: how to connect, the tools it offers, and the manifests MCP directories read.

Connect

  • Server: https://mcp.heycactus.ai/mcp
  • Transport: Streamable HTTP
  • Sign-in: OAuth 2.0 with your Cactus account (Google or GitHub). Dynamic client registration and client ID metadata documents are supported, with PKCE (S256).

Step-by-step setup for each AI client: https://heycactus.ai/docs/connect-an-ai-client

In Claude Code:

sh
claude mcp add --transport http cactus https://mcp.heycactus.ai/mcp

In Cursor, add this to .cursor/mcp.json:

json
{  "mcpServers": {    "cactus": {      "url": "https://mcp.heycactus.ai/mcp"    }  }}

You need a Cactus team with a product built from a GitHub repository. Cactus is about to launch; its closed beta started on 16 Oct 2026. Join the waitlist at https://heycactus.ai.

Tools

On https://mcp.heycactus.ai/mcp, signed in:

ToolWhat it does
list_productsList your team's products
search_helpSearch the product's published help pages
read_pageRead a help page
list_topicsList help topics
read_timelineRead the product's dated record of releases
lookup_mapRead the product map Cactus keeps from your code
search_conversationsSearch past customer conversations
read_conversationRead a conversation
search_evidenceSearch past customer conversations and imported help articles
search, fetchSearch, and fetch a result
askAsk Cactus a question. It answers from the published pages, the product map, the product knowledge and past conversations, and cites its sources
start_investigation, get_investigationInvestigate a question in the product's code and knowledge, and read the finding

Every tool carries MCP annotations. No tool edits or deletes anything. Only ask and start_investigation aren't read-only, because each runs a new answer or investigation.

Public help-center servers

Every product's public help center has its own server, with no sign-in: https://mcp.heycactus.ai/p/<name>/mcp. Cactus's own help center is https://mcp.heycactus.ai/p/cactus/mcp. It offers search_help, read_page, list_topics, read_timeline, ask_help, contact_team, search and fetch.

ask_help answers the way the help center answers a visitor. It starts from the published help pages and reads the product map, the code index and the code when those fall short. contact_team sends the team a message they can reply to.

Links

The files in this repository are under the MIT license.

Source: README.md at commit a3ad01b

Tools

0
This server asks you to sign in, so SourceWeft could not read its tools without an account and has not indexed them. What a connected account sees can depend on that account and its permissions.

Version history

1
  1. v1.0.0LatestOct 16, 2026

Score

85% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 daysRequired25 pts
  • Auth declaredRequired10 pts
  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts