Agree on what to build before any code is written.

Keep the source of truth in specs, and keep changes explicit.

This Setup Will Configure

• Agent Skills for AI tools
• Workflow-driven methodology and the current ApeWorkflow command surface
• Project-local workflow files

Workflow-Driven Methodology

ApeWorkflow is workflow-driven, not just command-driven.

Work moves through a clear path of workflow stages, and each stage carries the right methodology for the job. The visible command surface stays small on purpose, while the methodology skills keep the workflow consistent from stage to stage:

• Think through the problem with explore
• Turn intent into a change with propose
• Execute the work with apply
• Check the result with verify
• Close it out with archive

The methodology skills stay available as the internal playbook for those stages, so the workflow stays consistent instead of becoming a loose collection of prompts.

Why ApeWorkflow

ApeWorkflow gives AI-assisted development a shared contract:

• Align before implementation - capture intent in proposals and specs first
• Stay explicit - keep proposed work in apeworkflow/changes/ and current truth in apeworkflow/specs/
• Adapt as you learn - edit any artifact at any time, without forcing rigid phases
• Work across tools - generate native commands where supported, or use shared AGENTS.md guidance everywhere else

How It Works

flowchart LR
  E["/ape:explore"] --> P["/ape:propose"]
  P --> A["/ape:apply"]
  A --> V["/ape:verify"]
  V --> R["/ape:archive"]

  O["/ape:onboard"] -. "guided walkthrough" .-> P
  B["/ape:bulk-archive"] -. "archive many changes" .-> R
  F["/ape:feedback"] -. "send feedback" .-> R

The visible command surface is:

Core workflow: /ape:explore -> /ape:propose -> /ape:apply -> /ape:archive
Supporting commands: /ape:verify, /ape:onboard, /ape:bulk-archive, /ape:feedback

What ApeWorkflow Creates

After apeworkflow init, your project gets a local workspace like this:

apeworkflow/
├── specs/              # Source of truth
│   └── <domain>/
│       └── spec.md
├── changes/            # Proposed work
│   └── <change-name>/
│       ├── proposal.md
│       ├── design.md
│       ├── tasks.md    # Plain task outline
│       ├── plans/      # Execution plan files
│       └── specs/
│           └── <domain>/
│               └── spec.md
└── config.yaml         # Optional project configuration

Quick Start

1. Prerequisite

Node.js 20.19.0 or higher

2. Install

npm install -g @0xzace/apeworkflow@latest

Other package managers:

pnpm add -g @0xzace/apeworkflow@latest
yarn global add @0xzace/apeworkflow@latest
bun add -g @0xzace/apeworkflow@latest

Nix:

nix run github:0xzace/ApeWorkflow -- init

3. Initialize Your Project

cd your-project
apeworkflow init

4. Start the Workflow

/ape:onboard

Or jump straight to a change:

/ape:propose <what-you-want-to-build>

If you want to refresh generated instructions after setup, run:

apeworkflow config profile
apeworkflow update

Supported AI Tools

ApeWorkflow supports 30+ AI coding assistants, including:

Claude Code, Cursor, Codex, GitHub Copilot, Gemini CLI, Windsurf, Cline, Continue, OpenCode, Qwen Code, RooCode, Kilo Code, Kiro, Auggie, Amazon Q Developer, and more.

See the full list in Supported Tools.

CLI Highlights

| Command | Purpose | | --- | --- | | apeworkflow init | Initialize ApeWorkflow in a project | | apeworkflow update | Refresh generated instruction files | | apeworkflow list | List changes or specs | | apeworkflow view | Open the interactive dashboard | | apeworkflow show | Show a change or spec | | apeworkflow validate | Validate changes and specs | | apeworkflow archive | Archive a completed change | | apeworkflow status | Show artifact completion status | | apeworkflow instructions | Output enriched artifact instructions | | apeworkflow templates | Show resolved template paths | | apeworkflow schemas | List available workflow schemas |

Full reference: docs/cli.md

Docs

• Getting Started
• Workflows
• Commands
• CLI
• Supported Tools
• Concepts
• Customization
• Installation
• APE Workflow

Development

pnpm install
pnpm run build
pnpm test
pnpm run dev
pnpm run dev:cli

Telemetry

Anonymous usage stats are collected for command names and version only. No arguments, paths, content, or PII are collected. Telemetry is disabled in CI.

Opt out:

export APEWORKFLOW_TELEMETRY=0
export DO_NOT_TRACK=1

License

MIT