JSingletonAI

JSingletonAI / dejavu

Public

Memory that follows you across every AI tool. No cloud storage. No account required. Set it up once, use it everywhere.

deja-vu.dev ai-agents claude llm local-first mcp
83
6
89% credibility
Found May 20, 2026 at 83 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Deja Vu is a local-first AI memory layer for agents and assistants that stores context in SQLite on the user's machine, providing Python, REST, CLI, and MCP interfaces with privacy-focused Venice AI integration for LLM calls.

Star Growth

See how this repo grew from 83 to 83 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

What is Deja Vu?

Deja Vu is a local-first memory layer for AI tools. It solves the "AI tools forget everything between sessions" problem by storing your preferences, context, and learned facts in SQLite on your machine. You add memories once through the Python SDK, CLI, REST API, or MCP, and any connected AI tool can retrieve them. It uses Venice AI for the actual LLM calls that extract and rank what matters, but everything else runs locally. No accounts, no cloud, no telemetry by default.

Why is it gaining traction?

The privacy story is strong: memories live in a SQLite file you can open and inspect yourself. No vendor lock-in. The "one memory store, every tool" pitch is compelling -- context from one AI assistant is immediately available in another. The multi-interface approach (Python SDK, CLI with agent mode, REST server, MCP server) means you can start simple with the CLI and scale into custom agents without switching memory systems. The agent mode output is explicitly designed for tool loops, which shows they understand the target audience.

Who should use this?

Developers tired of re-explaining their preferences to every AI tool. Teams wanting persistent memory without cloud dependencies. Anyone with strict data residency requirements who wants AI context stored locally. The Python SDK works well for building custom agents; the CLI is useful for quick manual adds and searches; MCP support means it slots into Claude Desktop, Cursor, and similar tools.

Verdict

Early and unproven. The 83 stars and 0.8999999761581421% credibility score tell you this is a small, new project without community validation. The architecture is sound and the local-first approach fills a real need, but at v0.1.0 with minimal test coverage and evolving APIs, production use is premature. Worth experimenting with for personal projects or evaluating in non-critical workflows -- just don't bet production on it yet.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.