cold-start

Import user data sources in a consent-gated sequence with ClawVisor credential access.

1|Updated May 9, 2026
One-click install
npx skills add https://github.com/weiping/gbrain-cn --skill cold-start-weiping
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: cold-start
Source: https://github.com/weiping/gbrain-cn/tree/main/skills/cold-start
Command: npx skills add https://github.com/weiping/gbrain-cn --skill cold-start-weiping

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Cold-start data bootstrapping turns an empty gbrain into a live, queryable knowledge base by importing the highest-leverage sources in the right order while preventing unsafe credential handling.

Core Features & Use Cases

  • Sequenced day-one imports: Guides the user through phased ingestion of markdown/Obsidian, Google Contacts, Google Calendar (last 90 days), Gmail (signal sampling), conversation exports, X/Twitter archive, file archives, and meeting transcripts.
  • Safety-first credential handling with ClawVisor: Ensures API access happens through ClawVisor so the agent never holds raw OAuth tokens or API keys, with explicit consent gating for each phase.
  • Entity cross-linking during import: Detects entities and updates/links people, companies, meetings, and timelines as it imports so the brain compounds over time rather than staying isolated.
  • Resumable progress tracking: Maintains state in ~/.gbrain/cold-start-state.json so interrupted sessions can resume without repeating earlier work.

Quick Start

Activate the brain by telling the agent to run the cold-start skill and import your day-one sources with consent gating enabled.

Frequently Asked Questions about cold-start

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I bootstrap an empty knowledge graph with existing data sources?

Bootstrapping an empty knowledge graph requires importing high-leverage sources like markdown vaults, Google Contacts, and calendar data in a sequenced, consent-gated ingestion process to create a queryable base.

Can I import Google Calendar and Gmail data without exposing raw API credentials?

Yes, importing Google Calendar and Gmail data uses ClawVisor-mediated credential access, ensuring the agent never holds raw OAuth tokens or API keys while sampling signal data.

How do I link entities like people and meetings during a data bootstrapping session?

Entity cross-linking happens automatically during data bootstrapping, detecting and updating people, companies, and timelines as markdown files and conversation exports are imported.

Does day-one data bootstrapping support resuming interrupted Gmail or file archive imports?

Day-one data bootstrapping supports resumable execution by maintaining progress state in a local JSON file, allowing interrupted Gmail or file archive imports to resume without repeating work.

What is the best way to import an Obsidian vault and X/Twitter archive into a new brain?

The best way to import an Obsidian vault and X/Twitter archive is through a phased day-one setup workflow that safely sequences ingestion and applies entity detection to cross-link imported pages.

What happens if I interrupt a meeting transcript and contact import session?

If you interrupt a meeting transcript and contact import session, resumable progress tracking saves state locally, allowing you to resume the bootstrapping sequence exactly where it stopped.