Fde Consultant Skill

README

# FDE Consultant Skill

**Forward Deployed Engineering methodology, installable on any AI coding agent.**

The Skill teaches agents to apply the 4-stage FDE loop (Scoping → Prototyping → Production → Feedback) with a verifiable Trust Score, instead of generating generic AI advice.

## Install (1 command)

```bash
curl -fsSL https://raw.githubusercontent.com/selectess/fde-consultant-skill/main/install.sh | bash
```

Compatible with **Claude Code**, **Cursor**, **Windsurf**, **Codex CLI**, **OpenClaw**, **Continue.dev**, **ChatGPT**, **Gemini**.

**Zero install** (web LLMs): copy-paste `skill/ZERO-INSTALL.md` into ChatGPT or Claude.ai.

## What you get

- **Skill methodology** (`skill/SKILL.md`): 14 Operating Principles + 10 Anti-Patterns
- **6 MCP tools** (stdio transport): `fde_decompose`, `fde_roi`, `fde_scientific_search`, `fde_evals`, `fde_ontology`, `fde_trust_score`
- **3 plugin manifests**: Claude Code, OpenAI Codex, OpenClaw
- **4 deliverable templates** with mandatory `## Trust Score` section
- **8 Python scripts** (zero-dependency, stdlib only)
- **11 references** (deep knowledge, loaded on demand)
- **7 examples** + **3 case studies** with Trust Score 90-93
- **Modex multi-agent**: 4 roles + orchestrator + 3 shell scripts
- **Modex Plugin $6 lifetime**: ed25519 license + Stripe webhook + plugin loader
- **Trust Score registry**: 4 entries with SHA-256 hash chain (append-only, git-backed)
- **Independent Certifier**: re-derives the Skill's Trust Score from raw evidence

## Trust Score of this Skill

**100/100 Certified** by an independent Certifier role (`modex/certify_skill.py`). See `SKILL_TRUST_SCORE.json` for the certification artifact.

The Trust Score is human-computable, verifiable by anyone with `shasum`, and requires no SaaS dependency.

## Verify

```bash
git clone https://github.com/selectess/fde-consultant-skill.git
cd fde-consultant-skill
python3 -m pytest skill/tests/ modex/tests/ -q
python3 -m modex.certify_skill --skill-path ./skill --output ./cert.json
```

**Expected**: 75/75 tests pass, Trust Score 100/100 Certified.

## Architecture

3-layer model:
- **Layer 1**: Interface (Claude Code, Cursor, Windsurf, Codex, OpenClaw, Continue.dev, ChatGPT, Gemini)
- **Layer 2**: MCP server (6 tools via stdio)
- **Layer 3**: Core logic (8 scripts + 4 templates + 11 references + 7 examples + 3 case studies)

## Products

| # | Product | License | Price | Status |
|---|---|---|---|---|
| 1 | **FDE Skill** | Apache-2.0 | Free | Ship-ready (75/75 tests) |
| 2 | **MCP Modex** (1-agent local) | MIT | Free | Ship-ready |
| 3 | **Modex Plugin** | BSL | $6 lifetime | Ship-ready |
| 4 | **Modex Collective Plugin** | BSL | $260 lifetime | Waitlist V1.2 |
| 5 | **MCP Cloud + Registry** | BSL | $99-499/month | Waitlist V1.2 |
| 6 | **Enterprise Services** | Custom | $10K-100K | Waitlist V1.2 |

## License

Apache-2.0.