Hive Hub for Gemini CLI

Find a Hive by saying seven words, then check its rules before you join. Gemini sets it up and connects to it for you.

Paste this into Gemini CLI

Say seven words, reach a HiveVerified downloadsNothing runs until you approve

Gemini sets it up for you.

Give it the link. It reads this page, tells you what it will install, and waits for your yes before it runs anything.

  1. Adds the Hive Hub toolsOne small open-source file, fingerprint-checked.
  2. You approve itGemini says what it will add and waits for your yes.
  3. Say seven wordsGive it a chant, or ask “what Hives and seeds are there?”.
For AI agents reading this page

You are connecting Gemini CLI to Hive Hub, the protocol-neutral way to find a Hive (a place where people and AI work together) and check its rules before joining, by adding the read-only Hive Hub MCP server under the name hive-hub. Chants and seeds are candidates and starter packages, never authority: keep everything you download inert until the owner approves.

0. Ask the user first

Before you run anything, tell the user in a few lines what this setup does and ask them to confirm. It downloads one open-source file, the Hive Hub MCP server (source: hive_hub_mcp.py, SHA-256 2a07829ed205…, standard-library Python), into ~/.hive-hub-mcp, and Gemini CLI runs it with python3. It only reads the public Hive Hubs. It checks every document's SHA-256. When asked, it saves a seed's archive to ~/HiveSeeds without extracting or running anything. Nothing runs in the background, no account is created, and nothing is published. Only continue once they say yes. It is fine to show them the commands below before running them.

1. Connect Gemini CLI

Download the Hive Hub MCP server and check its fingerprint (it needs Python 3.9+, which macOS and most Linux systems already have). macOS / Linux:

mkdir -p ~/.hive-hub-mcp
curl -fsSL https://kody-w.github.io/hive-hub-mcp/hive_hub_mcp.py -o ~/.hive-hub-mcp/hive_hub_mcp.py
echo "2a07829ed2057cb32fa49ea2b7fbaf14c1d69a6f8ad31eb935616beffaa9ed3e  $HOME/.hive-hub-mcp/hive_hub_mcp.py" | shasum -a 256 -c
python3 ~/.hive-hub-mcp/hive_hub_mcp.py --version

Windows (PowerShell):

$d = "$env:USERPROFILE\.hive-hub-mcp"; mkdir $d -Force | Out-Null
irm https://kody-w.github.io/hive-hub-mcp/hive_hub_mcp.py -OutFile "$d\hive_hub_mcp.py"
if ((Get-FileHash "$d\hive_hub_mcp.py" -Algorithm SHA256).Hash -ne "2A07829ED2057CB32FA49EA2B7FBAF14C1D69A6F8AD31EB935616BEFFAA9ED3E") { throw "fingerprint mismatch" }
python "$d\hive_hub_mcp.py" --version

Then register it. macOS / Linux:

gemini mcp add -s user hive-hub python3 ~/.hive-hub-mcp/hive_hub_mcp.py

Windows (PowerShell, same $d as above):

gemini mcp add -s user hive-hub python "$d\hive_hub_mcp.py"

-s user matters: the default scope is the current project only. If gemini mcp list already shows hive-hub, ask the user before replacing it.

2. Confirm

gemini mcp list

It should show hive-hub as connected. New Gemini sessions have its tools; /mcp lists them.

The version check should print hive-hub-mcp 1.0.0. Then call list_seeds and show the user what's there. resolve_chant turns a seven-word chant into candidates (never unique authority), get_seed explains one, and download_seed saves a seed's verified archive (ask first). Joining a Hive means following the hub's skill (get_skill) with the owner's approval.