JSPM

@jondavis23/deep-context

0.1.1
  • ESM via JSPM
  • ES Module Entrypoint
  • Export Map
  • Keywords
  • License
  • Repository URL
  • TypeScript Types
  • README
  • Created
  • Published
  • Downloads 6
  • Score
    100M100P100Q63157F
  • License MIT

CLI-first AI coding assistant with persistent memory - your AI learns from corrections and remembers project decisions

Package Exports

  • @jondavis23/deep-context
  • @jondavis23/deep-context/dist/index.js

This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (@jondavis23/deep-context) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.

Readme

Deep Context

Your AI keeps forgetting what you told it. Deep Context fixes that.

The problem

Monday: "Use tabs, not spaces" Tuesday: "I said tabs!" Wednesday: "WHY ARE THERE SPACES"

Sound familiar? Every new chat starts from zero. Your AI has no memory.

What this does

Deep Context gives AI tools memory. You tell it something once, it remembers forever.

Install

npm install -g @jondavis23/deep-context

Setup

cd your-project
dc init
dc install

dc init detects your stack and sets up rules. dc install connects it to your AI tools (Claude Code, Cursor, Claude Desktop).

After that, your rules and preferences are automatically included in every AI conversation for that project.

You can add your own rules too:

dc add rule "Always use async/await, never callbacks"
dc add rule "All IDs must be UUIDs"
dc add choice "PostgreSQL" --why "Need complex queries"

How it works

Each project gets a .dc folder with:

  • Your rules (things the AI should always do)
  • Your choices (decisions you made and why)
  • Memories (things the AI learned while working with you)

When an AI tool asks for context, Deep Context gives it the relevant stuff. The AI writes better code because it knows your preferences.

Commands

dc init              # set up a project
dc install           # connect to Claude Code, Cursor, etc.
dc status            # see what's connected
dc add rule "..."    # add a rule
dc add choice "..."  # record a decision
dc disable           # turn off for this project
dc enable            # turn back on
dc remove            # delete everything

Does it actually work?

We ran a benchmark with 10 coding tasks. Without Deep Context, Claude followed project conventions 33% of the time. With Deep Context, 91%.

That's the difference between "use UUIDs" being ignored vs being followed.

Privacy

Everything stays on your machine in the .dc folder. Nothing is sent anywhere unless you're using a cloud AI (and even then, only the context goes to the AI you chose).

Requirements

  • Node 20+
  • For the MCP server: Claude Code, Cursor, or Claude Desktop

Questions

Do I need to configure anything? Run dc init then dc install. It auto-detects your project type and connects to your AI tools.

What if I have 50 projects? Each project has its own .dc folder. No bleeding between projects.

Can I turn it off for a project? dc disable creates a .dcignore file. The AI tools will skip that project.

Is it free? Yes. MIT license.

MIT License