Skip to content
Get packs

Guide · Agent skill

How to use the Mubu Live Bridge CLI agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

This guide explains how to install and use the cli-anything-mubu skill, a set of instructions that teaches an AI agent to run a command-line tool connected to the Mubu outlining app on your own computer. It is written for people who have never used agent skills. By the end you will know what the skill lets the agent do, how to install it, how to ask for things in plain language, and which safety rules to keep in place.

Mubu Live Bridge CLIHKUDS/CLI-Anything · Apache-2.0

Quick start

  1. 1Confirm prerequisites. Check that Python 3.10 or newer is installed and that the Mubu desktop app is running with an active session on this computer.
  2. 2Install the skill. Download the skill ZIP from this page and unzip it into ~/.claude/skills/ for personal use or a project's .claude/skills/ folder, or upload the ZIP under Settings, Capabilities, Skills in the Claude apps.
  3. 3Install the CLI. From the agent-harness source tree, run pip install -e . so the cli-anything-mubu command is available.
  4. 4Set your daily folder. Optionally set the MUBU_DAILY_FOLDER environment variable so daily helpers work without an argument.
  5. 5Ask for read-only results first. Start with a request to find or search content, which the skill handles through discover and inspect commands.
  6. 6Approve edits after reviewing the dry run. For any change, check the dry-run payload the agent shows you, then allow the --execute run only if it is correct.

What an agent skill is and what this one does

An agent skill is a folder with a SKILL.md file that tells an AI agent how to perform a task with specific tools. This skill, from the HKUDS/CLI-Anything project (agent-harness source, Apache-2.0 licence), teaches the agent to use a command-line interface for Mubu.

With the skill in place, the agent can list folders and documents, search note content, find your current daily note and inspect its nodes. It can also create, update or delete individual nodes in a live Mubu document, but those changes are previewed first and applied only when the command is explicitly run with an execute flag.

  • Discover: list folders, documents, recent documents and daily documents
  • Inspect: show document trees, search note text, list live nodes
  • Mutate: create a child node, update node text or delete a node, dry-run by default
  • Session: store the current document and node so later commands reuse them

Check the requirements before you install

The skill needs Python 3.10 or newer, and an active Mubu desktop session on the same machine. The CLI reads local Mubu profile data, so it will not work without that data being available. The source also lists an optional environment variable, MUBU_DAILY_FOLDER, for daily helpers that run without an argument.

The skill runs scripts and uses the network: the CLI reads local storage and calls the live Mubu API.

  • Python 3.10+
  • An active Mubu desktop session on this machine
  • Local Mubu profile data accessible to the CLI
  • Optional: MUBU_DAILY_FOLDER set to your daily folder reference

Install the skill for Claude

Option one is a personal install. Download the skill ZIP from this page, unzip it, and place the folder in ~/.claude/skills/. Option two is a project install: put the folder in a project's .claude/skills/ directory so it applies only to that project.

Option three is to upload the ZIP in the Claude apps under Settings, then Capabilities, then Skills.

The CLI itself is installed from the agent-harness source tree with pip. The source gives this command, run from that tree:

pip install -e .

Run the CLI and set your daily folder

After installation the CLI can be started with either command below. Run without a subcommand, it opens an interactive session that keeps document and node context between commands.

If you want the daily helpers to run without typing a folder reference, set MUBU_DAILY_FOLDER first. Otherwise pass the daily folder reference explicitly.

  • Use 'help' inside the REPL to see built-in commands
  • Use session commands to store current-doc and current-node
cli-anything-mubu python -m cli_anything.mubu

Ask the agent for things in plain language

You do not need to type commands yourself. Describe the outcome you want and the agent picks the right discovery or inspection command. The skill's recommended order is to find the daily document, inspect its nodes, store the document as the current one, then run a mutation only after checking the payload.

Ask for read-only work first. Requests such as the ones below are the kind the skill is built for. For any change, ask the agent to show the dry-run payload before it adds --execute.

  • Find my current daily note and list its top-level nodes.
  • Search my Mubu notes for anything mentioning the quarterly budget.
  • List the documents in my Projects folder and show the most recent ones.
  • Show me the node ids under the heading 'Action items' in today's daily note.
  • Draft a change to the text of that node, show me the dry-run, and wait for my approval before executing.

Review and approve live edits

Mutation commands are dry-run by default and only apply a change with --execute. A dry-run shows the exact outgoing payload, which is the point at which you should check it. The skill tells agents to prefer node ids over text matching, and to verify the result after any live mutation.

Keep the safety limits in mind. delete-node removes the whole targeted subtree, and even an update with the same text can advance the document's version history.

  • Prefer --node-id and --parent-node-id over text matching
  • Use --json whenever an agent will parse the output
  • Deleting a node deletes everything beneath it
  • Back up important documents before letting an agent edit them
cli-anything-mubu mutate update-text '<doc-ref>' --node-id <node-id> --text 'new text' --json

What people use it for

  • Asking an agent to find today's daily note and summarise its top-level items
  • Searching across Mubu backups for a phrase you half remember
  • Listing documents in one folder or folder path to plan a week of notes
  • Building a script that reads Mubu documents as structured JSON output
  • Preparing a targeted text edit to one node, reviewed as a dry run before it is applied

Frequently asked questions

How do I install the cli-anything-mubu agent skill?

Download the skill ZIP from this page and unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. You can also upload the ZIP in the Claude apps under Settings, then Capabilities, then Skills.

What do I need to run the Mubu CLI skill?

You need Python 3.10 or newer, an active Mubu desktop session on the same machine, and local Mubu profile data the CLI can read. The CLI is installed from the agent-harness source tree with pip install -e .

Is the Mubu live bridge skill free to use?

The source lists the licence as Apache-2.0. This page does not state any price, so check the licence terms in the source repository for details.

Can the agent change my Mubu notes without asking?

Mutations are dry-run by default and only apply when the command is run with --execute. Deletions remove whole subtrees and edits can advance version history, so review the payload before approving any live change.

Why does the Mubu CLI not work on my computer?

The tool depends on a running Mubu desktop session and local Mubu profile data on the same machine. If either is missing, the CLI cannot read your documents. Check that Mubu is open and signed in, and that Python 3.10 or newer is installed.

How does this compare with other Mubu integrations?

The source does not compare this skill with other tools, so this page cannot make that claim. Its distinguishing features are the JSON-based interface, the dry-run-first live edits and the agent workflow it describes.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Choose this skill if you use Mubu on a desktop computer and want an AI assistant to browse, search and inspect your notes, or if you are building structured automation and will review every edit before it runs. Pick something else if you want unattended editing of important documents, or if your environment forbids agents from touching local app data or calling remote APIs.

Download the skill