FerroxLabs

FerroxLabs / ijfw

Public

IJFW — It Just F*cking Works. Ferrox Labs' local-first infrastructure for AI coding agents: shared memory, smart routing, multi-AI cross-audits, disciplined workflow.

ferroxlabs.com
20
4
100% credibility
Found May 27, 2026 at 20 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
JavaScript
AI Summary

IJFW (It Just F*cking Works!) is a local-first coordination layer that connects multiple AI coding assistants so they share one persistent memory, route work to the right model, and cross-check each other's output. One install unifies up to 13 different AI platforms (Claude Code, Codex, Gemini, Cursor, Windsurf, Copilot, and others) with a shared brain that survives restarts and remembers project decisions across sessions. The Trident feature runs three AI models in parallel to catch blind spots before production. Everything stays entirely on the user's machine — no cloud account, no telemetry, no subscription. Costs are reduced through aggressive caching, smart routing, output trimming, and skill hot-loading. The project is open source under MIT license.

How It Works

1
🔍 Someone hears about IJFW

A developer hears from a colleague that there's a tool that makes all their AI coding assistants work together as one brain, sharing memory and catching each other's mistakes.

2
⚡ One command to install everything

They run a simple install command, and within seconds their computer has IJFW configured across all their AI coding tools without any complex setup.

3
🧠 Their AI starts remembering things

The first time they use any AI assistant, it instantly knows about their project. Decisions made last week are remembered. No more repeating yourself to each AI.

4
🔄 Three AI models review the same work

When they want extra confidence, they ask for a cross-audit. Three different AI systems each review the same code and share what they found — blind spots covered, quality up, costs tracked.

5
Solo developer or team project
👤
Solo work

A focused workflow keeps every session on track with clear phases and gates — no more mid-session drift or forgotten plans.

👥
Team project

Specialist AI agents are generated for their exact project needs and coordinate together, with memory that survives across every session.

6
📊 They see their actual savings

A simple dashboard shows exactly how much the AI is costing, how much context window is left, and what the memory system saved them by not repeating work.

✅ Everything stays local

Their code and memories never leave their machine unless they explicitly ask. They got a sophisticated AI coordination system that respects their privacy while making every AI assistant smarter and more efficient.

Sign up to see the full architecture

5 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 ijfw?

IJFW is a local-first infrastructure layer that connects your AI coding agents into one coordinated brain. It runs entirely on your machine, requires no accounts or cloud services, and makes Claude Code, Codex, Gemini, Cursor, Windsurf, Copilot, and a dozen other tools share the same memory and workflow discipline. The core is JavaScript with an MCP server handling the memory layer and bash hooks managing session lifecycle.

Why is it gaining traction?

The hook is simple: one install, and your AI stops being amnesiac. IJFW persists decisions, patterns, and handoffs across sessions so you never restart from scratch. The Multi-AI Trident feature runs Codex and Gemini in parallel to audit code from different AI lineages, catching blind spots that a single model misses. Token savings are real and measurable: prompt caching, smart routing between Haiku/Sonnet/Opus, output discipline, and command sandboxing compound into a lower bill that the dashboard tracks per session. The preflight command runs 11 quality gates in under 90 seconds before you ship.

Who should use this?

Developers who work across multiple AI coding tools and are tired of context loss between sessions. Teams running AI-assisted projects where quality gates and audit trails matter. Solo developers who want their AI to remember project conventions without manual re-explaining. If you use Claude Code or Codex professionally and care about token costs or code quality, this addresses pain points you probably already feel.

Verdict

IJFW solves real problems with a thoughtful architecture, but 20 stars and a 1.0% credibility score mean you are early. The local-only, zero-dependency design is architecturally sound, and the feature depth is impressive for this stage. Worth evaluating if you want to standardize your AI workflow, but treat it as a cutting-edge tool that may evolve rapidly.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.