PrePilot MCP Server
Enables AI clients to compose and execute marketing campaigns by mapping free-form intent to a deterministic plan of over 1,000 production-tested skills, supporting multiple workflows across paid ads, SEO, content, and more.
README
PrePilot — AI Marketing Agency Suite
<div align="center">
MCP-first, multi-agent orchestration platform for marketing agencies.
Turn intent into a deterministic skill composition plan — across 1,000+ marketing skills, in any language, for any workflow.
Website · Live Demo · Documentation
</div>
✨ What is PrePilot?
PrePilot is an open-source marketing agency in a box. It exposes a deterministic orchestration layer that maps free-form marketing intent (Arabic, English, mixed) to a runnable plan — composed from a catalog of 1,000+ production-tested skills, wired through the Model Context Protocol (MCP) so any AI client (Claude Code, Cursor, Custom GPT, Codex, Gemini CLI) can plug in and ship campaigns end-to-end.
┌─────────────────────────────────────────────────────────────────┐
│ AI Client (Claude Code · Cursor · Custom GPT · Codex · Gemini) │
└──────────────────────────────┬──────────────────────────────────┘
│ MCP / SSE / Custom GPT Actions
▼
┌──────────────────────────────────────────────────────────────────┐
│ PrePilot MCP Server │
│ ─────────────────────────────────────────────────────────────── │
│ • Intent router → 13 canonical workflows (paid, SEO, ...) │
│ • Skill catalog → 1,040 skills mapped to taxonomy │
│ • Swarm compiler → agents.py + manifest.json + workflow │
│ • Engagement engine→ role → artifact → role → artifact ... │
└──────────────────────────────┬──────────────────────────────────┘
│
▼
┌──────────────────────────────────────────────────────────────────┐
│ Compiled Agent Runtime │
│ • 1,700+ pre-built agents (worker · coordinator · critic) │
│ • 4 topology modes (pipeline · parallel · hierarchical) │
│ • Append-only memory (rules · episodes · agency data) │
└──────────────────────────────────────────────────────────────────┘
🚀 Quick Start
git clone https://github.com/imMamdouhaboammar/PrePilot-oss.git
cd PrePilot-oss
pip install -e ".[dev]"
# Start the sample MCP server
python examples/mcp-server/server.py
# Wire it into Claude Code
claude mcp add prepilot-sample -- python examples/mcp-server/server.py
Then in your AI client:
You: Launch a paid Meta campaign for my DTC skincare brand in Saudi.
PrePilot: I'll run the `launch_paid_ad_campaign` workflow.
[step 1/6] audience research (geo-fundamentals + market-segments)
[step 2/6] brand voice capture (brand-voice-guide)
...
[step 6/6] ad creative generation (ad-copy + ad-creative-brief)
✓ Plan ready. Estimated cost: SAR 12,000.
🎯 Who is PrePilot for?
| You are... | PrePilot helps you... |
|---|---|
| Marketing agency owner | Run 10× the campaigns with the same headcount |
| Brand marketer | Get a senior strategist + executor pair on demand |
| Marketing ops engineer | Stop building one-off campaign scripts — compose from tested skills |
| Indie hacker | Ship a full GTM playbook in an afternoon |
| AI client developer | Plug into 1,000+ production skills without writing them yourself |
📦 What's in this repo
PrePilot-oss/
├── README.md ← you are here
├── LICENSE ← MIT
├── CONTRIBUTING.md ← how to add your skill
├── CODE_OF_CONDUCT.md
├── docs/
│ ├── ARCHITECTURE.md ← how PrePilot works end-to-end
│ ├── MCP_INTEGRATION.md ← wire into Claude / Cursor / GPT
│ ├── SKILLS_GUIDE.md ← anatomy of a skill
│ └── ROADMAP.md ← what's next
├── examples/
│ ├── mcp-server/ ← runnable MCP server skeleton
│ ├── chatkit-widgets/ ← sample ChatKit widgets for GPT apps
│ └── skills/ ← 25 sample skills (canonical taxonomy)
├── src/
│ └── compiled_agents/ ← 25 sample agents (worker + coordinator)
├── templates/
│ ├── SKILL.template.md
│ ├── WORKFLOW.template.md
│ ├── PROMPTS.template.md
│ ├── skill.yaml
│ ├── input.schema.json
│ └── output.schema.json
├── assets/
│ └── (logos, diagrams)
├── pyproject.toml
└── .github/ ← issue templates + CI workflows
🧠 Architecture in 30 seconds
- User intent arrives ("launch paid Meta campaign in Saudi")
- Intent router classifies it:
language=en,workstream=paid_ad,complexity=multi_step - System layer recommends the matching workflow (
launch_paid_ad_campaign) - Workflow materializes into ordered steps with skill IDs injected
- Each skill compiles into an
agent.py+manifest.json(deterministic codegen, no LLM in the loop) - Swarm executes in the chosen topology (pipeline / parallel / hierarchical / DAG)
- Engagement engine validates each artifact contract before handoff to the next role
- Memory accumulates — episodic archive, rule inbox, agency decisions
For the deep dive, see docs/ARCHITECTURE.md.
🔌 MCP Integration
PrePilot is MCP-native. Any MCP-compatible client (Claude Code, Cursor, Codex, Gemini CLI, ChatGPT with MCP support, etc.) can plug in directly.
| Transport | Use when |
|---|---|
| stdio | Local AI client (Claude Code on your machine) — recommended |
| SSE/HTTP | Remote AI client (Cloud Run deploy, ChatGPT widget, public MCP gateway) |
| Custom GPT Actions | OpenAI Apps SDK with the OpenAPI schema exported from this repo |
Full wiring instructions in docs/MCP_INTEGRATION.md.
🧩 Skills Catalog (this repo ships 25 of 1,040)
The full catalog lives in the private production deployment. This OSS repo ships 25 representative skills spanning every major marketing domain — enough to compose any workflow and to demonstrate the skill authoring pattern.
| Domain | Sample skill |
|---|---|
| Paid acquisition | ad-copy, ad-creative-brief, landing-page-architect |
| SEO / AEO / GEO | seo-technical, geo-fundamentals, programmatic-seo-content |
| Brand & voice | brand-voice-guide, value-prop-statements, ambassador-program |
| Content & media | content-strategy, content-ops, podcast-ops |
| Email & lifecycle | email-nurture-sequence |
| Analytics & growth | ab-test-analysis, linkedin-engager-analytics, cash-flow-forecast |
| Strategy & ops | competitor-analysis, market-segments, fulfillment-sop, compliance-checklist |
| Sales & conversion | sales-script, growth-loops |
| Copywriting | copywriting |
| Chat & messaging | telegram-mini-app |
Want to add your own skill? See CONTRIBUTING.md and the templates/ folder.
🤝 Contributing
We welcome:
- 🆕 New skills — drop a folder under
examples/skills/<your-skill-id>/withSKILL.md,skill.yaml,input/output.schema.json - 🐛 Bug reports — open an issue
- 📖 Docs improvements — PRs welcome
- 🌍 Translations — Arabic, Spanish, French, German, etc.
See CONTRIBUTING.md for the full guide.
📜 License
This project is licensed under the MIT License — see LICENSE.
🔐 Production vs OSS
The OSS release is a curated, public-safe subset of the full production system. The production deployment includes:
- 1,040 production skills (vs 25 here)
- 1,700+ compiled agents (vs 25 here)
- Proprietary orchestration SOPs (paid-campaign-launch, content-program, brand-refresh, ppc-audit)
- Live ad-platform integrations (Meta, GA4, Google Ads)
- Customer-tuned agency memory (brand voices, campaign histories, decisions)
- Cost & pacing optimization engines
The OSS version is designed to be fully runnable and to demonstrate the architecture — without exposing the proprietary IP that powers the production SaaS.
🌐 Links
- 🌐 Website: prepilot-system-agency.space
- 📦 Production app: app.prepilot-system-agency.space
- 💬 Discord: [coming soon]
- 🐦 Twitter / X: @imMamdouhaboammar
<div align="center">
Built with ❤️ in Cairo · Shipped to the world
</div>
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.