ce-ai

CE-AI — Cross-Host Operational Companion for Compound Engineering

Cross-host operational companion for Compound Engineering across AI coding agents. Compound Engineering owns engineering semantics and workflow artifacts, hosts own native execution and packaging, and ce-ai coordinates the environment around them: fleet version governance, native installation orchestration, project adoption, and advisory workflow observation without replacing coding agent reasoning.

Try it in two minutes

Prerequisite: install one supported coding-agent host first.

# Linux / Unix (Standalone scripts, Go & Cargo)
curl -fsSL https://raw.githubusercontent.com/mastepanoski/ce-ai/main/scripts/install.sh | bash
curl -fsSL https://raw.githubusercontent.com/colbymchenry/codegraph/main/install.sh | sh
go install github.com/Gentleman-Programming/engram/cmd/engram@latest && cargo install rtk

# Windows (PowerShell, Go & Cargo)
irm https://raw.githubusercontent.com/mastepanoski/ce-ai/main/scripts/install.ps1 | iex
irm https://raw.githubusercontent.com/colbymchenry/codegraph/main/install.ps1 | iex
go install github.com/Gentleman-Programming/engram/cmd/engram@latest; cargo install rtk

# macOS / Homebrew
brew install mastepanoski/ce-ai/ce-ai gentleman-programming/tap/engram rtk
curl -fsSL https://raw.githubusercontent.com/colbymchenry/codegraph/main/install.sh | sh

# In your project: initialize graph, adopt project, govern fleet, and verify
cd your-project && codegraph init
ce-ai init-prj && ce-ai fleet pin latest && ce-ai fleet sync && ce-ai doctor

Then reopen your coding agent in your-project and start with /ce-brainstorm <outcome>. For package manager alternatives and detailed steps, see Getting Started.

Documentation map

Document	Audience	Intent
🌱 Getting Started	Beginner	Tutorial — install, adopt a project, and run a first workflow step
🎓 CE-AI positioning	Beginner / Senior	Explanation — agents, Compound Engineering, fast-path, and CE-AI’s boundary
🚀 Quick Start Workflow Guide	Beginner	Tutorial — first feature, bug fix, and workflow resumption
🎓 Compound Workflow Explained	Beginner	Explanation — how strategy becomes code and durable learning
🎓 Documentation Debt & Hygiene	Beginner	Explanation — keep specifications and guidance current
📁 Project Adoption Guide	Both	How-to — safely adopt or de-adopt a project
🎓 Harnesses, Loops & Context	Beginner	Explanation — coding-agent hosts, context, and supporting tools
🎓 Determinism Explained	Beginner	Explanation — what CE-AI can and cannot guarantee
🎓 Zero-Step Drift Recovery	Beginner	Explanation — recover current repository state at session start
🛡️ Agent Guardrails Guide	Both	How-to — command risk checks, security guardrails, and readiness
🔧 Installation & Coexistence	Both	How-to — install without overwriting existing configurations
🔄 Fleet, Sync & Upgrade	Both	How-to — multi-harness fleet governance, reconcile drift, and batch maintenance
⚡ Skill Registry Guide	Both	How-to — find and resolve available skills
💾 Backup & Uninstall	Both	How-to — restore configurations and remove CE-AI safely
🗂️ Harness Matrix	Senior	Reference — supported hosts, configuration paths, and integration methods
🏛️ Architecture Guide	Senior	Explanation — deterministic state, adapters, and project artifacts
🎮 FSM & Checkpoints · vs. Memory	Both	Explanation — lifecycle stages, checkpoints, and session memory
🎓 Architecture vs. Chat TDD	Senior	Explanation — context economics, workflow granularity, and bounded TDD
📖 CLI reference · 🏛️ Living Architecture	Senior	Reference — current commands, boundaries, and living contracts
🧠 Solutions Library · Plans & Audits	Contributor	Reference — solved problems, decisions, and delivery history

Acknowledgments

CE-AI builds on EveryInc’s Compound Engineering. Early exploration of agile fast-paths drew inspiration from Alan Buscaglia’s Gentle AI. It also draws on Engram, CodeGraph, Context7, RTK, and Sequential Thinking.

Project links

Security · AI policy · Contributing · Documentation style · MIT License