loom

Store and search project decisions in a SQLite-backed memory store.

Updated Aug 23, 2026
One-click install
npx skills add https://github.com/NeimadLab/claude-skills --skill loom-neimadlab
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: loom
Source: https://github.com/NeimadLab/claude-skills/tree/main/skills/loom
Command: npx skills add https://github.com/NeimadLab/claude-skills --skill loom-neimadlab

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Loom provides persistent, structured project memory that survives across AI tools and sessions, enabling cross-model handoffs, searchable decisions, and reliable context transfer.

Core Features & Use Cases

  • Persist and search decisions, goals, risks, and notes across Claude Code, Cursor, Windsurf, and Claude.ai.
  • Generate structured handoff summaries and compact onboarding context for new sessions or models.
  • Apply governance via a policy engine and audit trail to keep remote and local actions observable and controllable.

Quick Start

Initialize a Loom workspace in your project, then connect your AI tool and start logging decisions with loom_log_decision.

Frequently Asked Questions about loom

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

FAQPage Schema
How do I persist project memory and decisions across different AI tools?

You can persist project memory across different AI tools by using a SQLite-backed memory store with an MCP gateway to log, search, and transfer structured decisions and context across sessions.

Can I transfer context and generate handoffs between Claude Code, Cursor, and Windsurf?

Yes, you can transfer context and generate structured handoff summaries between Claude Code, Cursor, Windsurf, and Claude.ai by utilizing its cross-model session management and MCP gateway capabilities.

How does AI development governance and action policy enforcement work?

AI development governance works via a built-in policy engine and audit trail that observes and controls remote and local actions, ensuring that AI-assisted development processes remain reliable and controllable.

What is the best way to onboard a new AI model into an existing development project?

The best way to onboard a new AI model is to generate compact onboarding context from previously logged decisions and goals, allowing the new session to immediately access searchable project history.

Do I need SQLite to manage persistent memory for AI-assisted development?

You need SQLite because it serves as the underlying backing store for the memory persistence layer, enabling structured search and retrieval of project notes, risks, and goals across tools.

Why does cross-tool context transfer fail without structured memory persistence?

Cross-tool context transfer fails without structured memory persistence because individual AI sessions lack a shared state, making historical decisions, goals, and audit trails inaccessible to new models.