M
MACRO TECH TITAN
The Engine of Macro Tech Titan
def maximus():
async def run():
np.dot(w, x)
while True:
@cached
self.power+=1
def maximus():
async def run():
np.dot(w, x)
while True:
@cached
self.power+=1
def maximus():

Maximus 🐴

Maximus 🐴

Maximus 🐴

The Workhorse forFintech

Not the intelligence — the architecture. You bring the intelligence. Maximus brings the engine: the skills, tools, memory, and tireless execution to act on it.

Free · Open · Ungated, forever · Runs on any model, or a free local one

Maximus is an AI-powered engine — context, skills, tools, code, plugins, connectors, disciplines, rules, and hard-won experience, packed into one workhorse. Built for builders, power users, programmers, developers, designers, entrepreneurs, small teams, and the enterprise alike.

Read the manifesto: The future of AI is the orchestration layer
xAI
Perplexity
Anthropic
Automate
Research
Generate
Maximus AI — the Tennessee-built workhorse
Get started →View on GitHub
maximus / live stats
Lines of code
…
github.com/MacroTechTitan/MaximusAI
Use cases live
5
and growing
GitHub stars
…
open · MIT
Tech in the engine
PythonTypeScriptNodeReactFastAPIPostgresRedisDockerOllamaOpenAI-compatible
The workhorse kit

What turns a model into a workhorse

Maximus sits on top of OpenClaw — an open-source, local-first assistant gateway. On that foundation it bolts on the four parts that turn a raw model into something that actually pulls weight:

Skills

Encoded procedures for hauling specific loads — gotchas baked in. Deeper skills, heavier loads.

Tools / MCP

Real hands on GitHub, databases, deploy platforms, the web. A horse with no harness pulls nothing.

Context

Keeps the thread of your intent across long, multi-step work.

Memory

Remembers so you never repeat yourself. This is what makes Maximus yours, not a rental.

Built for technical people — engineers, builders, founders, scientists — using the tools they already ship with. Not a retail chatbot. No hand-holding, no friction you didn't ask for.

A power tool for people who know how to use power tools — rigorous, resourceful, dry-humored, and allergic to hand-waving.

The Maximus Suite

42 skills. 5 pillars. One workhorse.

Each skill is a self-contained instruction set an AI agent loads on demand. No dead weight, no hand-waving. Built for engineers, founders, and scientists.

Skills load only when their triggers match. You carry no dead weight.

Cognitive OS· 1 skill

maximus-brain

Cognitive operating system: think-before-act, memory hygiene, depth-adaptive execution. The meta-skill that sits above the rest.

View on GitHub →

Build & Ship· 10 skills

maximus-design-spec

Write a software design spec before code.

View on GitHub →

maximus-plan-implementation

Break an approved design into a minimum-change plan.

View on GitHub →

maximus-build-feature

Implement features with read-before-edit discipline.

View on GitHub →

maximus-code-review

Review diffs and PRs for correctness, security, performance.

View on GitHub →

maximus-debug-incident

Reproduce, isolate, hypothesize, fix, regress.

View on GitHub →

maximus-eval-and-test

Unit, integration, e2e, and LLM evals.

View on GitHub →

maximus-devops-ship

CI/CD, IaC, progressive delivery, rollbacks.

View on GitHub →

maximus-fintech-payments

Stripe, webhooks, idempotency, PCI awareness.

View on GitHub →

maximus-python-scientific

Reproducible pipelines: pinned deps, fixed seeds.

View on GitHub →

maximus-replit-handoff-pro

Production handoffs between Computer and Replit Agent.

View on GitHub →

AI Engineering· 15 skills

maximus-agent-design

Tool loops, memory, recovery, 3-tier evals.

View on GitHub →

maximus-prompt-engineering

Production system prompts, JSON schemas, few-shot.

View on GitHub →

maximus-rag-pipeline

Chunking, embeddings, hybrid search, reranking, citation grounding.

View on GitHub →

maximus-llm-model-selection

Pick the right model for cost/latency/quality.

View on GitHub →

maximus-ai-product-spec

Behavior, evals, staged rollout, kill switch.

View on GitHub →

maximus-ai-safety-governance

PII redaction, injection defense, audit logs.

View on GitHub →

maximus-ai-data-pipeline

Dataset curation, labeling, DVC, dataset cards.

View on GitHub →

maximus-fine-tuning

When to fine-tune vs RAG vs prompt; LoRA, QLoRA, DPO.

View on GitHub →

maximus-ai-fluency-for-builders

The meta-skill for delegating well.

View on GitHub →

maximus-ai-cost-control

Token economics, prompt caching, $/request budgets.

View on GitHub →

maximus-mlops-deploy

Model registry, canary, drift, rollback.

View on GitHub →

maximus-ai-ux-patterns

Streaming, citations, confidence, guardrails.

View on GitHub →

maximus-k3-model-selection

New

Decide when Kimi K3 wins vs. Claude Fable 5 / GPT-5.6 Sol / Opus 4.8 / GLM-5.2. Benchmark-cited, honest, refuses to recommend K3 when fit is worse.

View on GitHub →

maximus-k3-self-hosting

Plan and execute a self-hosted K3 deployment on vLLM / SGLang / TokenSpeed with MXFP4 weights, preserved-thinking, and a license gate.

View on GitHub →

maximus-chain-of-verification

New

Apply factored CoVe as the final layer before delivery: draft → independent verification questions → fresh-context answers → revise. 40–60% documented hallucination reduction. Never labels a claim verified without independent-context confirmation.

View on GitHub →

Writing, Research & People-Finding· 9 skills

maximus-write-article

Long-form articles: thought leadership + technical builds.

View on GitHub →

maximus-deep-research

Multi-source synthesis and competitive intelligence.

View on GitHub →

maximus-deep-research-pro

Inference-driven research: hypothesis-first, adversarial verification, confidence ledger.

View on GitHub →

maximus-investigative-research

New

Research the way a working journalist or intelligence analyst does it: source tiers (primary/secondary/adversarial), timeline construction, contradiction hunt, corroboration pass. Produces a story with a spine, not a topic summary.

View on GitHub →

maximus-literature-review

New

Systematic literature review with PRISMA-style flow, inclusion/exclusion criteria, quality appraisal (RoB 2 / NOS / ROBINS-I / field-adapted), and GRADE confidence per finding.

View on GitHub →

maximus-people-finder

Deep 7-step agent for investors, journalists, partners, board members, experts.

View on GitHub →

maximus-people-finder-recruiter

Deep 8-step recruiter agent for candidate sourcing.

View on GitHub →

maximus-counterparty-discovery

Finance-grade counterparty discovery: SEC filings, court dockets, source-by-source provenance, compliance gate before outreach.

View on GitHub →

maximus-contact-intelligence

Find the most likely professional business email from a LinkedIn URL, with pattern discovery, verification, and confidence scoring. Never labels a guess as verified.

View on GitHub →

AI SEO Pack — opt-in· 7 skills

Opt-in pack

Get cited by ChatGPT, Perplexity, Google AI Overviews, and Copilot. Answer Engine Optimization + Generative Engine Optimization + classical technical SEO, unified.

maximus-ai-seo-strategy

The strategy layer.

View on GitHub →

maximus-aeo-optimization

Answer Engine Optimization for LLM answers.

View on GitHub →

maximus-geo-optimization

Generative Engine Optimization for Perplexity, SGE, Copilot.

View on GitHub →

maximus-technical-seo

Core Web Vitals, JSON-LD, crawlability, JS SEO.

View on GitHub →

maximus-content-seo

On-page, internal linking, E-E-A-T, refresh cadence.

View on GitHub →

maximus-seo-audit

Umbrella audit that ties the pack together.

View on GitHub →

maximus-llm-visibility-tracking

Measure citation share across Perplexity, ChatGPT, Claude, Gemini, AI Overviews.

View on GitHub →
View the AI SEO pack overview →

Try the workhorse.

Open the repo →Read the launch post
FREE ✦OPEN ✦UNGATED ✦FOREVER ✦ANY MODEL ✦LOCAL-FIRST ✦SKILLS ✦TOOLS ✦MEMORY ✦EXECUTION ✦THE ENGINE ✦NOT THE MIND ✦FREE ✦OPEN ✦UNGATED ✦FOREVER ✦ANY MODEL ✦LOCAL-FIRST ✦SKILLS ✦TOOLS ✦MEMORY ✦EXECUTION ✦THE ENGINE ✦NOT THE MIND ✦
What Maximus IS

An engine, not a model

Every company building in AI sells a model — intelligence for rent. Macro Tech Titan builds an engine.

Maximus isn't the intelligence; you are. It's the architecture that turns what a sharp human already decided into work that actually gets done — the skills, the tools, the memory, the execution.

It plows your digital fields, harvests the crops, ships them to store. It doesn't think for you. You are the rider.

The free-engine model

Free engine. Build anything on it.

🔓

Forever ungated

The engine is free, open, and ungated — forever. Not a free tier, not a trial. The backbone itself never costs anything.

🌾

We sell what it helps you grow

We build products on Maximus — AI Connect, trading signals, and more — and those products can charge for convenience, hosting, and premium use-cases. The engine that powers them is always free.

We don't sell the horse. We sell what the horse helps you grow.

Two paths

Pick your saddle

🐴

Run your own Maximus

A free, local-first engine that does the toil. Bring your own model — or run a free local one with no keys and no bill. Set it up in minutes, teach it about you, put it to work.

→ Quickstart
🛠

Build on Maximus

Open-source skills, packs, and deployable agents. Compose what you need, write your own, or drop a standalone agent into your stack. MIT licensed, free forever.

→ Developer docs · Browse the repo
What we believe

The rider decides

AI is an engine, not a mind.

It executes; it does not replace human judgment.

The human is in command.

Not consulted — in command. Maximus acts on intent; it doesn't substitute its own.

Capability over cleverness.

A plain answer that ships the work beats a brilliant one that doesn't.

Free is a feature, not a tier.

The engine is ungated by design and forever.

Built on Maximus

The engine behind everything

Maximus is the engine behind everything Macro Tech Titan builds. The products charge for the use-case. The engine underneath stays free.

AI Connect

Build-and-ship orchestration and DevOps platform.

OptimaQuant

Algorithmic trading signals and strategy marketplace.

Maximus Content Engine AI

Generate, schedule, and ship content at scale — powered by the engine.

Deal Match AI

AI-matched deal flow for buyers, sellers, and intermediaries.

…and more

More products in the works, all powered by the same free engine.

Quickstart

Talk to Maximus in minutes

Requires OpenClaw (Node 22.19+ or 24).

bash
git clone https://github.com/MacroTechTitan/MaximusAI.git
cd MaximusAI
./install.sh

# 1. Edit memory/memory.md to teach Maximus about you
# 2. Copy config/openclaw.example.json → ~/.openclaw/openclaw.json
#    Pick a free local model (Ollama, no keys) or bring your own API key
# 3. Start the gateway
openclaw gateway

Talk to it through OpenClaw's WebChat or any connected channel — Telegram, Slack, Discord, Signal, and more. No custom UI required.

Agents — standalone apps

Deployable agents, same philosophy

Full-stack agent apps you can clone whole or drop a single agents/<name>/ directory into your own stack. The skill is a procedure Maximus follows locally; the agent is a hosted implementation you deploy and visit.

recruiter-deep-find

Recruiter deep-find

Recruiter intake form + live deep-search. Sources candidates from DevOps agencies, dev shops, outsourcing firms, MSPs, and technical staffing agencies.

React · Vite · Express · Perplexity Sonar

find-shareholders

Find shareholders

Input company names (public or private); get institutional shareholders (funds) and named individuals at those funds. Exports CSVs shaped for Folk and Apollo enrichment.

React · Vite · Express · Perplexity Sonar

Why it's free — honestly

The Maximus layer is free forever: persona, skills, memory, tooling. The one cost no open project can erase is model inference — compute is never free. So you choose: a local model on your own hardware (truly free, no keys), or your own API key (frontier quality, you pay only your own inference). Maximus never gates a feature behind money or routes you through a paid middleman.

For developers

MaximusAI/
├── core/     SOUL.md · AGENTS.md · TOOLS.md
├── skills/   one folder per capability
├── packs/    opt-in domain bundles
├── memory/   persistent context store
├── mcp/      external tool wiring
├── config/   example OpenClaw config
├── install.sh
├── CLAUDE.md build brief for agent-driven dev
└── agents/   deployable agent apps

The core is generic. Domain-specific work lives in a pack/ you opt into. Compose, don't fork. MIT licensed.