Fde Skill

README

# FDE Consultant

Open-source Forward Deployed Engineering methodology for AI coding agents and personal agents.

FDE Consultant helps agent runtimes such as Claude Code, Codex, Cursor, Windsurf, Hermes, and open-source agent stacks behave less like generic code generators and more like field operators: scope first, prototype with evals, productionize deliberately, and productize what they learn.

## Launch Positioning

- **Open-source skill first:** the public product lives in `skill/` and is free to inspect, install, and extend.
- **Academy as community layer:** `marketing/landing/academy.html` teaches the FDE method for free.
- **Blog as growth engine:** the blog applies Hybrid Entity Pulse to rank the project around FDE, AI scoping, evals, handoffs, and agent skills.
- **MCP Beta later:** `mcp-server/` is the hosted MCP/SaaS foundation, but it should be presented as Beta/coming soon until production hardening is complete.

## Repository Map

| Path | Purpose |
|------|---------|
| `skill/` | The FDE Skill (Apache-2.0). 14 Operating Principles, 10 Anti-Patterns, 6-trait eval rubric, 8 Python scripts, 7 examples, 4 deliverable templates, 3 case studies with Trust Scores 90-93. |
| `modex/` | The Modex multi-agent orchestrator (BSL). 4 roles (Lead/Researcher/Builder/Certifier), init.sh + verify.sh, hash-chain registry. |
| `mcp-server/` | The hosted MCP server (BSL, Bêta). 6 MCP tools, FastAPI, Stripe billing scaffolding. Not production-ready yet. |
| `registry/` | Public append-only Trust Score registry. SHA-256 hash chain, verified by `skill/tests/test_registry_chain.py`. |
| `marketing/landing/` | The public website: index, docs, academy, blog hub, 24 blog articles. |
| `work/` | Working notes, scoping reports, audits. |

## Quick Start

### Option 1: Zero-Install (Web UI: ChatGPT, Claude.ai)
Copy-paste [skill/ZERO-INSTALL.md](skill/ZERO-INSTALL.md) into the system prompt.

### Option 2: Extra High Level Installer (CLI Agents: Claude Code, Cursor, Windsurf, Codex)
```bash
curl -fsSL https://raw.githubusercontent.com/selectess/fde-skill/main/install.sh | bash
```

### Option 3: Manual Install
See [skill/INSTALL.md](skill/INSTALL.md).

## Why FDE Consultant?

The market has a massive gap in AI agent methodology. 43% of AI projects fail. Palantir charges $400K-$600K/year for Forward Deployed Engineering — 99% of companies cannot pay. The `SKILL.md` format has become an open standard in 2026.

FDE Consultant is the open-source answer: **the FDE methodology every AI agent should run.**

## License

- `skill/` — Apache-2.0
- `modex/`, `mcp-server/` — BSL (Business Source License)
- Documentation — CC-BY-4.0