luml-ai

luml-ai / dreamer

Public

Self-evolving context for any coding agent.

17
1
100% credibility
Found May 06, 2026 at 17 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Dreamer is a self-hostable service that pools memories from team coding agents into shared, auto-updating context files.

How It Works

1
🔍 Discover Dreamer

You hear about Dreamer from a blog post or teammate—it's a way to make your team's AI coding helpers share what they learn across everyone.

2
📦 Set up your space

Run a simple command to create folders and a settings file for collecting team memories.

3
🔐 Create team access

Make a secure private code so your AI helpers can safely share their experiences.

4
Check and start

Quickly verify everything looks good, then launch the memory collector with one click.

5
🚀 Connect your helpers

Point your coding tools like Claude or Cursor to your new service—they start sending memories automatically.

6
Watch memories grow

On a schedule, Dreamer blends all the shared learnings into updated guides your team uses next time.

🎉 Smarter team agents

Your AI coding buddies now remember team lessons forever, working better together without starting over.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 17 to 17 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

# Dreamer

Dreamer pools short-term memories from team coding agents—Claude Code, Cursor, custom CLIs—into a shared long-term store, then dreams up evolving context like AGENTS.md and skills on a cron schedule. Self-hostable MCP server; agents submit via `submit_memory` tool over HTTP.

Standout

- Team memory: Single bundle from all agents, not per-workstation silos. Beats solo dreamer designs. - Plug-and-play: YAML wires SQLite STM, Markdown LTM/context, Claude Agent dreamer—swap for Postgres, graph DBs, custom engines. - MCP native: Any MCP client works; gates (budget, empty batch), hooks (git commit, GitHub PR), triggers (cron, threshold). - No dreamer lyrics or Ozzy vibes—this is practical agent infra, not PyTorch world models or scenario dreamers.

Part of LUML's open agent stack (core registry, Prisma research agents, Flow tracing).

Setup

`pip install 'dreamer-server[defaults]'` `dreamer init` (YAML + workspace) `dreamer-simple-auth token create` `dreamer serve`

CLI dreams one-shot, checks config compliance.

Verdict

Raw prototype for awesome self-evolving agents GitHub hunters: 17 stars, 1% credibility screams experiment. Protocols shine for extension—build your dreamers Chemnitz/Offenbach style—but skip for prod until battle-tested. Fork if lucid dreamer GitHub vibes hit; watch LUML for v1 polish.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.