// Who we are

Built to Own
the Full Stack.

Proverbs was born from a single conviction: developers should own their tools, their models, and their data. No subscriptions. No cloud dependencies. No permission needed to code.

"By wisdom a house is built, and through understanding it is established." — Proverbs 24:3

// The builder

RAXX BEATS STUDIOS LLC

Proverbs is designed and maintained by Shane Bedasee under RAXX BEATS STUDIOS LLC — a solo software studio building privacy-first, developer-sovereign tools.

SB
Shane Bedasee
// Founder · RAXX BEATS STUDIOS LLC
Full-stack developer and systems architect. Built Proverbs to eliminate the dependency on cloud coding services for day-to-day software development. Believes every developer deserves world-class tooling at zero recurring cost.
🏢
RAXX BEATS STUDIOS LLC
// Wellington, FL · Est. 2020
A software development company specializing in smart coding tooling, SaaS platforms, and developer infrastructure. All products are built with a privacy-first ethos — data stays where it belongs: with the user.
// Company details
CompanyRAXX BEATS STUDIOS LLC
LocationWellington, FL, USA
ProductProverbs Faith-Based Smart Coding Language
Websiteproverbscode.com
GitHubgithub.com/raxxbeats/proverbs
Contactstizzyraxx@gmail.com

// Why we exist

The Mission

The goal is straightforward: eliminate every external dependency in the coding workflow until the entire stack — from inference to indexing to tooling — runs on your machine.

mission.txt — proverbs
MISSION STATEMENT
"Trust in the LORD with all your heart and lean not on your own understanding."
— Proverbs 3:5  ·  Our work is the tool. The wisdom is yours.
# What we are building toward:

GOAL_1 = "No cloud inference — every token generated locally"
GOAL_2 = "No paid APIs — open-weight models only by default"
GOAL_3 = "No data telemetry — zero bytes leave the user's machine"
GOAL_4 = "No subscriptions — free forever, open source core"
GOAL_5 = "Full sovereignty — developer owns every layer"

# Current status:
✓ Local inference via Proverbs (own weights) or GGUF fallback — DONE
✓ Full tool suite offline — DONE
✓ Multi-backend routing (Ollama / LM Studio / Jan / llamafile / Built-in) — DONE
✓ Proverbs — custom engine built from scratch (tokenizer, attention, training loop, inference server) — DONE
✓ Ollama-free inference via llama-cpp-python GGUF backend — DONE
✓ Full local training pipeline (Mac CPU offline mode + Google Colab GPU mode) — DONE
✓ Custom inference server — node-llama-cpp, port 11435 — DONE
✓ Direct cloud APIs (Anthropic / OpenAI, no proxy) — DONE
✓ No-key web search (DDG + Bing HTML scraping) — DONE
✓ Multi-agent mode (/agents) — DONE
✓ Security auditor (/security) — DONE
✓ ONNX embeddings (no Ollama for vectors) — DONE
○ Bundled fine-tuned coding model — PLANNED

// How we think

Our Philosophy

Three principles guide every decision we make — from architecture choices to default settings to what features we'll never build.

🔒
Privacy-First
Your code is proprietary. When you use a local model, nothing you type, no file you open, and no output generated ever touches a remote server. The default answer to "does this leave my machine?" is always no. Cloud fallback is opt-in and clearly disclosed.
"Keep your heart with all vigilance, for from it flow the springs of life." — Proverbs 4:23
📡
Offline-First
Internet connectivity is a privilege, not a guarantee. Proverbs is designed to be fully functional on a plane, in a bunker, or behind a corporate firewall with zero egress. The network is a bonus, never a requirement.
"The prudent sees danger and hides himself." — Proverbs 22:3
⚔️
Developer Sovereignty
You should be able to read every line of code that processes your prompts, swap out any model, add any tool, and ship a custom build without asking permission. Open source core. No telemetry. No license server. No phone home.
"The beginning of wisdom is this: get wisdom, and whatever you get, get insight." — Proverbs 4:7
🌱
Radical Transparency
Every tool call Proverbs makes is printed to your terminal in real time. You see the exact file it reads, the exact diff it writes, the exact command it runs — before it runs. Nothing happens behind the scenes.
"Honest scales and balances belong to the LORD." — Proverbs 16:11
💰
Zero-Cost by Default
The default experience — running Proverbs locally — costs exactly $0 after hardware. No API key needed. No subscription needed. No free tier with artificial limits. The paid cloud fallback exists for when you need it, not as the primary path.
"A generous person will prosper; whoever refreshes others will be refreshed." — Proverbs 11:25
🛠️
Tools, Not Magic
Proverbs is a deterministic tool suite with a reasoning engine. When it writes a file, you see the diff. When it runs a command, you see the output. There is no "trust us" — you verify every action yourself.
"Plans are established by seeking advice." — Proverbs 20:18

// Where we're going

Roadmap to Full Independence

Three phases to eliminating every external dependency from the developer workflow.

✓
// Phase 1 — Complete
Multi-Backend Router
Proverbs runs its own inference server as the primary backend. Also supports LM Studio, Jan, and llamafile as fallbacks. No config changes required. Direct Anthropic and OpenAI API calls replace the LiteLLM proxy for cloud fallback. Use /backend to inspect or force a backend at runtime.
Ollama LM Studio Jan llamafile Built-in server Direct Anthropic API Direct OpenAI API 55+ slash commands
✓
// Phase 2 — Complete
Custom Inference Server
Proverbs runs as your primary inference backend on port 11434. Built from scratch in PyTorch — engine architecture, BPE tokenizer, training pipeline. GGUF fallback via llama-cpp-python for instant use before your own model is trained.
node-llama-cpp Proverbs server (Ollama-compatible API) OpenAI-compatible API GGUF model support /server start|stop|status Auto-starts on boot
Full Independence
The final layer: ship Proverbs with everything self-contained. Web search already uses direct HTML scraping (no API key). Cloud fallback calls Anthropic/OpenAI directly (no LiteLLM proxy). Remaining: ONNX embeddings (no Ollama for vectors), whisper.cpp auto-bundle (no system whisper install), and a fine-tuned coding model trained on open-source repos — zero third-party dependencies from install to inference.
No-key web search Direct cloud APIs ONNX embeddings whisper.cpp auto-bundle Bundled fine-tune Zero-dependency installer
"Commit your work to the LORD, and your plans will be established." — Proverbs 16:3