You will get a custom MCP server connecting Claude or Cursor to your codebase and APIs
Project details
You get a custom MCP server that connects Claude, Cursor, or any LLM client to your codebase, database, and internal APIs through the Model Context Protocol. No more rebuilding the same agent scaffolding or pasting context by hand.
I build the server in TypeScript, define each tool to match your stack, add tests so you know when something breaks, and write setup docs your team can follow. I have shipped a production MCP server published on npm and several AI tools running in real products.
I start with a short spec and push back when a request will cost more than it returns. You receive working, documented code, deployed to your environment if needed. I work async in clear written English.
I build the server in TypeScript, define each tool to match your stack, add tests so you know when something breaks, and write setup docs your team can follow. I have shipped a production MCP server published on npm and several AI tools running in real products.
I start with a short spec and push back when a request will cost more than it returns. You receive working, documented code, deployed to your environment if needed. I work async in clear written English.
AI Algorithms
Large Language Model, Multimodal Large Language Model, Transformer ModelAI Applications
AI-Generated Code, Conversational AI, Natural Language Generation, Natural Language UnderstandingAI Development Language
PythonAI Tools
GitHub Copilot, Hugging FaceAI Models
ChatGPT, GPT-4What's included
| Service Tiers |
Starter
$150
|
Standard
$450
|
Advanced
$900
|
|---|---|---|---|
| Delivery Time | 5 days | 10 days | 14 days |
Number of Revisions | 1 | 2 | 3 |
AI Model Integration | - | - | |
Batch Normalization | - | - | - |
Database Integration | - | - | |
Detailed Code Comments | - | - | |
Image Upscaling | - | - | - |
MLOps | - | - | |
Model Deployment | - | - | |
Model Documentation | - | - | |
Model Monitoring | - | - | - |
Model Testing & Optimization | - | - | - |
Model Tuning | - | - | - |
Natural Language Processing | - | - | - |
NLP Tokenization | - | - | - |
Pre-Training | - | - | - |
Prompt Engineering | - | - | |
Setup File | - | - | |
Source Code | - | - |
Frequently asked questions
About Abo
AI Agent & MCP Developer | RAG Systems & Production AI Tooling
Marrakesh, Morocco - 7:17 am local time
If your AI features feel like demos, your RAG pipeline hallucinates, or your team is rebuilding the same agent scaffolding for the third time, you need someone who has shipped this in production. That is what I do.
What I actually build:
MCP servers and AI tooling. Custom Model Context Protocol servers that let Claude, Cursor, and other LLM clients talk to your codebase, your database, your internal APIs. I have shipped a revenue-intelligence MCP server published on npm, and a GitHub App that flags business-logic vulnerabilities in pull requests and posts a fix-ready report with remediation steps.
Agentic workflows. Multi-step agents with tool calls, validation pipelines, and human-in-the-loop checkpoints. Not chatbots. Real workers that produce auditable output.
RAG systems that survive production. I have built document AI processing Arabic and English legal, financial, and government documents on LightRAG with graph-based retrieval, hybrid search, and evaluation pipelines so you know when retrieval breaks before your users do.
Full-stack AI SaaS. Next.js, Node, PostgreSQL, Supabase, Prisma. End-to-end product delivery from schema to ship, including auth, billing (Paddle and other Stripe alternatives for global founders), and CI/CD.
Recent shipped work (full links and demos in the Portfolio section below):
Scan & Action. AI receipt and invoice processing with a decision engine, live in production with Paddle billing, Supabase auth, and a Capacitor Android build in testing.
Fixor. GitHub App that detects six business-logic vulnerability classes (auth bypass, missing admin checks, IDOR, environment-variable exposure, hardcoded secrets, unverified webhooks) in pull requests, posts a fix-ready report, and tracks detector scope in a versioned capability contract. Deployed on Railway with a Next.js dashboard.
ContextOps MCP. Published npm package that maps an unfamiliar TypeScript SaaS codebase to its entry points, risky files, and revenue-relevant smells for AI coding agents.
KnowFlow. Arabic and English document AI with paid Paddle subscriptions and pgvector retrieval, processing legal contracts and financial reports.
How I work:
I write a spec before I write code. I push back when a request will cost more than it returns. I ship in small, reviewable increments. I tell you when something will not work instead of building it twice.
I work in Arabic and French as primary languages. English is fully fluent in writing and async collaboration. If your project needs daily voice calls in English, I am not your best hire. If it needs a builder who delivers, we should talk.
Steps for completing your project
After purchasing the project, send requirements so Abo can start the project.
Delivery time starts when Abo receives requirements from you.
Abo works on your project following the steps below.
Revisions may occur after the delivery date.
Scope & spec
I confirm the tools, data sources, and target LLM client, then write a short spec before any code
Build the MCP server
I implement the server and its tools in TypeScript, wired to your codebase, database, or APIs