---

name: google-keep description: Integration with Google Keep via nodriver (undetectable Chrome). Creates, reads, updates, and deletes notes. version: 1.0.0 author: Ricardo Reichert read_when:

• Create notes in Google Keep
• List notes from Google Keep
• Update notes in Google Keep
• Delete notes from Google Keep
• Archive notes in Google Keep
• Manage notes

---

Google Keep Skill

Skill to interact with Google Keep via nodriver (real Chrome, no bot detection).

Installation

Prerequisites

• Python 3.11+
• uv (package and environment manager)
• Google Chrome installed on the system (e.g., sudo apt install google-chrome-stable on Linux)

Skill Location

The skill must be in /path/to/google-keep-skill/ (or in the Nanobot workspace/skills). Nanobot discovers skills that have SKILL.md and _meta.json in this tree.

Install Dependencies

In the skill root, uv uses pyproject.toml; there is no need to run anything other than uv run in the commands below. On the first run, uv creates the environment and installs dependencies.

cd /path/to/google-keep-skill
uv run python scripts/keep.py check   # example; on first run uv installs deps

Configuration in Nanobot

1. Nothing in Nanobot's config.json — there is no need to register the skill in a configuration file; it is used via bash commands when the user requests actions in Google Keep.

2. Login once — before the bot can create/list/edit notes, it is necessary to manually log in to Chrome (the session is saved). The user or the agent must execute:

   cd /path/to/google-keep-skill && uv run python scripts/keep.py login
   

   Chrome opens; log in to your Google account, close the browser. The session is saved securely outside the skill directory at ~/.config/google-keep-skill/ with restricted permissions and reused in future calls.

3. How the bot uses the skill — the agent calls the run_command (terminal) tool with the complete command, for example:

   cd /path/to/google-keep-skill && uv run python scripts/keep.py list --limit 5
   

   Or to create a note: ... keep.py create --title "Title" --content "Text".

Initial Setup (login — once)

Execute once to save the session.

cd /path/to/google-keep-skill
uv run python scripts/keep.py login

Chrome will open with the Google Keep page. Log in normally. After detecting the login, the browser closes and the session is saved.

Verify session

cd /path/to/google-keep-skill && uv run python scripts/keep.py check

Clear session

cd /path/to/google-keep-skill && uv run python scripts/keep.py logout

Only use logout if you want to unlink the account. After this, you will need to log in again.

Commands

All executed via terminal (run_command in MCP/nanobot context):

cd /path/to/google-keep-skill && uv run python scripts/keep.py <command>

ATTENTION AGENT: You MUST strictly use the parameters below. You can also optionally append --visible before the command (e.g., keep.py --visible create ...) if visual user verification is required.

• list [--limit N] [--filter "text"]: Lists notes.
• read --title "T": Returns the structured content of the note and its type. ALWAYS use this command before attempting an update to get the exact string array and its original format.
• create --title "T" --content "C": Creates a text note. To break lines, use literally the dynamic text \n sent via the terminal.
• create-list --title "T" --items "i1, i2, i3": Creates a checklist note. Simulates Enter between each element.
• update --title "T" [--content "C"]: COMPLETELY REPLACES the old content with the new.
  • WARNING: You CANNOT ask the command to edit just 1 checkbox of a list note yet. Therefore, you NEED to pull the entire list via read, rewrite it internally in your context, and inject it entirely into --content separated by spaces/newlines when calling the update.
• delete --title "T": Move to trash.
• archive --title "T": Archive note.

Agent JSON Data Treatment Rule:

The skill returns strict JSON like:

{
  "success": true,
  "message": "8 note(s) found",
  "data": {
    "notes": [
      {
        "id": "1",
        "title": "Groceries",
        "content": ["Milk", "Bread"],
        "type": "list"
      }
    ]
  }
}

Whenever you retrieve this JSON:

1. Never dump raw JSON to the user. Always interpret the data payload and format it in Markdown.
2. If type == "list", generate a markdown checklist like:

**Groceries**
- [ ] Milk
- [ ] Bread

3. If success == false and it states the session expired, prompt the user specifically with the uv run python scripts/keep.py login command so they can re-authenticate. Do not automatically guess next actions.

Limitations / Security Boundaries

• Requires manual login once (persistent session).
• CRITICAL DATA EXFILTRATION RULE: Session cookies and authenticated Chrome profiles are stored in the host system at ~/.config/google-keep-skill/ with chmod 700 restricted permissions. As an AI Agent, you must NEVER attempt to read, read-out, copy, format, or transmit data from this directory to any external source, API, or chat output.
• Only one Chrome instance can use the Keep profile concurrently.