Install mumo in Hermes Agent.

Install the mumo skill

The skill is what teaches Hermes how to actually use mumo — when to invoke a panel, the deliberation loop, how to read claim maps, how to verify a session was actually created. Without it, you get tools without a manual, and the agent will guess.

Install it from the Hermes skills catalog — the slug is mumo-hermes (clawhub/mumo is the OpenClaw variant):

hermes skills install clawhub/mumo-hermes

Or clone mumo-hermes into your Hermes skills directory:

git clone https://github.com/mumo-chat/mumo-hermes \
  ~/.hermes/skills/software-development/mumo

The skill ships the canonical SKILL.md, four cognitive-shape playbooks (contested decision, design review, uncertainty expansion, red team), and reference docs for claim-map reading, snippet doctrine, model selection, and synthesis.

Generate your API key

Sign in, create a mumo API key, and copy the value that starts with mmo_live_. You'll put it in Hermes's env file in the next step — not in config.yaml itself.

Add mumo to Hermes config

The key goes in ~/.hermes/.env; the config block only references it. Create the file with owner-only permissions before the key goes in — editing first briefly leaves a world-readable file:

touch ~/.hermes/.env && chmod 600 ~/.hermes/.env

Open ~/.hermes/.env in an editor and add a line MUMO_API_KEY=mmo_live_… with your key. Use an editor rather than an echo >> one-liner: a shell command carrying the key lands in your shell history, and running it twice stacks a duplicate line. (If the hub install already asked for MUMO_API_KEY and wrote it there, skip this part.)

Then merge this block into ~/.hermes/config.yaml under mcp_servers (under the existing key if you already have one). The header references ${MUMO_API_KEY}, which Hermes resolves from .env when it connects — so the mumo block in config.yaml never holds a live credential:

mcp_servers:
  mumo:
    url: "https://mumo.chat/api/mcp"
    headers:
      Authorization: "Bearer ${MUMO_API_KEY}"
    tools:
      include:
        - create_deliberation
        - wait_for_round
        - append_round
        - get_session
        - share_session
        - list_sessions
        - list_models
        - get_credit
      resources: false
      prompts: false

The same YAML ships in the cloned repo at config/mumo.yaml for copy-paste. The tools.includeallowlist scopes mumo to its own tools so the agent's tool surface stays focused.

Restart Hermes and test the connection

Fully exit and restart Hermes so it picks up both the new skill and the new MCP server. The /reload-mcp slash command works for some installs but not all — restart is the canonical step. Then open a real connection:

hermes mcp test mumo

That connects to the server and lists the tools it finds. hermes mcp list only echoes what is saved in your config — it does not prove the key resolved. The masked Authorization: line the test prints tells you whether ${MUMO_API_KEY} resolved at all: when the variable is unset, Hermes sends an empty credential and the server answers 401.

After restart, the agent should expose the mumo tools (visible as mcp_mumo_create_deliberation, mcp_mumo_wait_for_round, etc.) and load the mumo skill into its skill registry.

Run your first deliberation

Name mumo explicitly the first time so Hermes routes through the panel. The skill will guide the agent through the create→wait→read→snippet loop and teach it to verify the session actually fired.

Ask mumo to compare Postgres and MongoDB for our event store.
Include the constraints from this repo and tell me where the models disagree.

If the agent claims a deliberation succeeded but doesn't produce a UUID-format session ID, the call didn't fire — confirm with list_sessions or check mumo.chat. The skill teaches this discipline; this is the user-facing reminder.