Kizai's project-level AI Coding SOP toolkit for controlled agent workflow governance.
Project description
kz-ai-coding-sop
Kizai's own AI Coding SOP for controlled, agent-based software development.
中文定位:Kizai 的可控式 AI Coding 标准作业流程。
Controlled AI coding means the human defines the rules, the agent follows the workflow, and every change stays understandable, reviewable, and verifiable.
What This Is
KZ AI Coding SOP is Kizai's project-level SOP and governance toolkit for AI coding agents. Its repository is:
- KZ repository: https://github.com/Kizai/kz-ai-coding-sop
It initializes a project with shared instructions for Codex, Claude Code, Cursor, Antigravity, OpenClaw, and similar agents.
KZ SOP uses a Superpowers-first workflow: when Superpowers skills are available in the current agent harness, KZ SOP routes agents to those skills as the workflow baseline. KZ SOP does not own, copy, bundle, modify, or claim authorship of Superpowers.
Superpowers is an upstream skill/workflow project maintained by Jesse Vincent and Prime Radiant:
- Repository: https://github.com/obra/superpowers
- License: MIT
In short: KZ SOP is Kizai's own development process and repository-level rules; Superpowers is an upstream skill capability that KZ SOP can reference and route to when installed.
Package Status
KZ SOP is distributed through npm and PyPI:
- npm package:
kz-sop - PyPI package:
kz-ai-coding-sop - CLI command:
kz-sop
Public install commands work for versions already published on the corresponding package registry. For local development or pre-release verification, use the local commands below.
Quick Start After Publishing
Use npm or pnpm:
npx kz-sop init
pnpm dlx kz-sop init
Or use Python:
pip install kz-ai-coding-sop
kz-sop init
This initializes the current project with KZ SOP files.
Local Usage Before Publishing
Python can be tested immediately from this repository:
cd kz-ai-coding-sop
python3 -m kz_sop.cli skills list
python3 -m kz_sop.cli init
python3 -m kz_sop.cli doctor
To install the local Python package into your environment:
cd kz-ai-coding-sop
python3 -m pip install .
kz-sop init
For npm, install Node.js first, then run:
cd kz-ai-coding-sop
npm install
npm run build
node dist/cli.js init
You can also test the package before publishing:
npm pack
Commands
kz-sop init
kz-sop update
kz-sop doctor
kz-sop skills list
initgenerates KZ SOP files without overwriting existing files.updateis a safe non-overwriting sync in v0.1.x.doctorchecks required KZ files and reminds you to confirm Superpowers is installed in your agent harness.skills listlists recommended upstream Superpowers skills and any packaged KZ-owned skills.
Typical Project Setup
In a new project:
mkdir my-project
cd my-project
git init
npx kz-sop init
or:
mkdir my-project
cd my-project
git init
pip install kz-ai-coding-sop
kz-sop init
By default, kz-sop init keeps the generated SOP files local and low-noise by adding them to .gitignore:
# KZ AI Coding SOP generated files
.kz/
AGENTS.md
If you want to make KZ SOP a shared team contract, remove those ignore entries intentionally and commit the files.
Superpowers should be installed separately in each agent harness you use. KZ SOP does not auto-install third-party plugins.
Packaged Skills
KZ SOP v0.1.x includes a small curated set of packaged skills:
startup-pressure-test: pressure-test startup ideas before building.grill-me: stress-test a plan or design through focused questioning.karpathy-guidelines: reduce common LLM coding mistakes while writing, reviewing, or refactoring code.waza-router: route work to Waza-style engineering habits.ecc-operator-system: audit cross-harness AI agent workflow readiness.
These extend KZ SOP without changing the default Superpowers-first workflow. Attribution for adapted upstream material is listed in THIRD_PARTY_NOTICES.md.
Generated Files
AGENTS.md
.kz/
sop.config.json
.gitignore
The .gitignore update only adds generated SOP paths:
# KZ AI Coding SOP generated files
.kz/
AGENTS.md
Recommended First Prompt
Please follow KZ AI Coding SOP for this project.
First read AGENTS.md.
Do not write code immediately.
Start by reporting project understanding, stack detection, structure analysis, plan, relevant Superpowers skills, risks, and verification steps.
Release Checklist
Before publishing npm:
npm install
npm run build
npm publish
Before publishing PyPI:
python3 -m pip install build twine
python3 -m build
python3 -m twine upload dist/*
Recommended pre-release verification:
python3 -m compileall kz_sop
python3 -m pip wheel . --no-deps -w /tmp/kz-sop-wheel
Philosophy
AI is not a free-form developer. It is an engineering worker that must follow project rules, load the right workflow skills, preserve architecture, verify results, and explain its changes.
KZ SOP owns project governance. Superpowers owns the default agent workflow.
Documentation
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file kz_ai_coding_sop-0.1.4.tar.gz.
File metadata
- Download URL: kz_ai_coding_sop-0.1.4.tar.gz
- Upload date:
- Size: 24.1 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.11.12
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
749f798e3b026e7ee86f5ebaa3b5c44586275eca4bb02012ee756b7f302f5c42
|
|
| MD5 |
12415ae0cdd46829f5be51c1d51bb5cf
|
|
| BLAKE2b-256 |
806a689103bd76068dadbcdd8d657d0e77d21e980ed53f42a0b4d0707af0f6b1
|
File details
Details for the file kz_ai_coding_sop-0.1.4-py3-none-any.whl.
File metadata
- Download URL: kz_ai_coding_sop-0.1.4-py3-none-any.whl
- Upload date:
- Size: 27.2 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.11.12
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
d0e54b1044d141e7aa2644d9cd2bab254e9aaa95bcc3a6cf4df0fa11087a3d62
|
|
| MD5 |
1f0e409acee8a14c3fec48c4e2292be6
|
|
| BLAKE2b-256 |
ac8fc63e601417415df0b230d152383fe2c0015c47bf962cc203c80038062ecf
|