decisionnode

CLI + Local MCP - A shared structured memory store across Claude Code, Cursor, Windsurf, Antigravity, and every MCP client. Semantically queryable.

20
2
100% credibility
Found Apr 16, 2026 at 20 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
TypeScript
AI Summary

DecisionNode stores project decisions in a searchable memory layer that AI coding tools can query to maintain consistency.

How It Works

1
🔍 Discover DecisionNode

You hear about a friendly tool that helps AI coding helpers remember key project choices so work stays consistent.

2
📥 Set it up easily

Download it once and start using it right in your project folder.

3
🔗 Link your AI buddy

Connect it to your favorite coding assistant with a quick setup.

4
💡 Record your first choice

Save what you decided and why – it becomes smartly searchable forever.

5
🖥️ Explore the magic map

Open the colorful view to see decisions linking up like a living mind map.

🎉 Consistency unlocked

Your AI now recalls every decision, keeping projects perfectly aligned.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 20 to 20 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 DecisionNode?

DecisionNode is a TypeScript CLI tool that creates a local, shared memory store for structured project decisions—like "skip connection pooling here"—queryable via semantic search with Gemini embeddings. It exposes these as MCP tools for AI clients across Claude Code, Cursor, Windsurf, Antigravity, letting agents pull relevant memories explicitly before coding. Run `decide init` in any repo, add via `decide add`, search with `decide search "pooling"`, or visualize in a local web UI showing graphs and vector spaces.

Why is it gaining traction?

Unlike static CLAUDE.md files bloating prompts, it retrieves only relevant decisions via cosine similarity, with conflict detection blocking duplicates at 75% match. The live UI pulses nodes when AIs query, plus history tracking sources changes from CLI or MCP clients. Works as a local CLI agent across tools, npm-installable for GitHub Actions, Linux/Ubuntu/Windows workflows, or quick `gh` clones.

Who should use this?

Backend devs enforcing architecture choices across AI sessions in Cursor/Claude. Teams sharing decisions via global flags in multi-tool setups like Windsurf + Copilot. Solo coders needing persistent, searchable memories without cloud dependency—ideal for local LLM/CLI integrations on GitHub repos.

Verdict

Worth installing for MCP-heavy AI coding (npm i -g decisionnode), especially with its polished docs and VSCode extension. At 20 stars and 1.0% credibility, it's early—expect rough edges—but the CLI + UI combo delivers immediate value for decision persistence.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.