Super Dev CLI
A CLI that wires a research-to-delivery pipeline with documents and quality gates into Claude Code, Cursor, Codex and other agents
- 278 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- Writes rule, command and skill files into the project's agent configuration, and into the home directory if you choose the user scope
- Following this pipeline the agent writes code, runs checks and prepares delivery, so it gets broad access to the project
- Installs a set of Python dependencies, including a web server and network request libraries
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/super-devDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Writes rule, command and skill files into the project's agent configuration, and into the home directory if you choose the user scope
- Following this pipeline the agent writes code, runs checks and prepares delivery, so it gets broad access to the project
- Installs a set of Python dependencies, including a web server and network request libraries
This is third-party code. Review the repository files before installing.
What it does
Super Dev installs with one command, `uv tool install super-dev`, and running `super-dev` offers a terminal menu to tick the agents you use. The program writes rules, commands and skills for the chosen agents into the project, and the work then continues inside them. A request starts with `/super-dev` or `super-dev:` followed by the task. For a new project the pipeline goes through research, three documents (product description, architecture and interface design), document confirmation, a spec, frontend, preview confirmation, backend, quality checks and delivery. For an existing project it first captures the baseline and waits for confirmation. The modes are `new`, `evolve`, `variant`, `patch` and `resume`. After a break the agent continues from files in `.super-dev/` and `output/`. The set includes 11 expert roles injected into prompts by stage, a knowledge base of hundreds of files, 25 YAML validation rules, a red team that audits security, performance and architecture, and a design system with palettes and tokens. Results go into reports such as `proof-pack` and a release readiness check.
Who it is for. For developers and small teams who hand a whole project to agents and want documents, confirmation steps and quality checks.
Good fit when
- A project from scratch has to go from research and documents to backend and delivery under confirmation gates
- An existing product is being extended and you need to capture and agree its baseline before changes
- A defect needs fixing on a light path with reproduction notes, impact area and regression checks
- Work must survive closing the window: state lives in `.super-dev/SESSION_BRIEF.md` and `workflow-state.json`
- For a 30-minute hackathon there is the shortened `/super-dev-seeai` branch
Not a fit when
- You need an edit in one file: the full pipeline with three documents and confirmations is too heavy
- You want a neutral set without hard stages: the agent will wait for confirmations after the documents and preview
- You need an English interface and documentation: the main README and messages are in Chinese, with a separate English version
- You cannot change agent configuration in the project: the program writes rule and command files there
Example request
/super-dev Build an appointment booking service for a small clinic: research and three documents first, then wait for my confirmationLimitations
You need Python 3.10 or newer and the `uv` package manager. The program does not install agents, Node.js, Docker or databases. Signing in to agent accounts stays with you. The main README is in Chinese, with English in `README_EN.md`.
The program separates "connected" from "verified": written files do not yet mean the agent really runs the pipeline. After installation it creates a guide, `output/maintenance/host-onboard-smoke-*.md`, with the first phrase for the agent. Agent status is split into certified, compatible and experimental.
The pipeline deliberately stops and waits for your decision after the three documents and after the frontend preview. Each stage requests a lot of text from the agent, so token use is higher than with short prompts.
How to disable. `super-dev uninstall --dry-run` shows what would be removed, and `super-dev uninstall` removes the files injected into the agents. The package itself is removed with `uv tool uninstall super-dev`. Delete the `.super-dev/` and `output/` folders in the project by hand.
FAQ
How do I install Super Dev and connect it to Claude Code?
Run `uv tool install super-dev`, then `super-dev`. In the menu pick the agent with arrows, tick it with Space and press Enter. Afterwards the program shows how to start the pipeline: `/super-dev your task` for agents with slash commands and `super-dev: your task` for the rest.
Which agents are supported?
The author names 26 agents: 12 command-line ones, 9 editor-based and 5 desktop assistants, including Claude Code, Codex, Gemini CLI, Cursor, Windsurf and Cline. Each gets a certified, compatible or experimental status.
What are the three documents and why confirm them?
They are the product description, the architecture and the interface design. The agent waits for your confirmation of the documents before moving to the spec and code. It also shows the frontend preview and waits for a separate confirmation.
How do I continue yesterday's work?
Tell the agent to continue the current process or ask what the next step is. State is restored from `.super-dev/SESSION_BRIEF.md` and `.super-dev/workflow-state.json`; no terminal commands are needed.
How do I remove Super Dev from the agents?
First run `super-dev uninstall --dry-run` and read the list. Then `super-dev uninstall` removes the injected files.
Discussion
No questions or reviews yet.
Related
CLI toolsA configurator that reads your repo and sets up agents, skills, MCP servers and rules for its stack across several assistants
A skills library that gives coding agents a development process: brainstorming, planning, TDD, subagents and code review
A CLI that gives six coding agents project context, over 80 skills, verification and resumable tasks across sessions
The /improve skill: a strong model audits a repository and writes self-contained fix plans, and you hand the execution to a cheaper model
A Claude Code plugin that extracts behavioral specs, test vectors and acceptance criteria from a codebase, with a provenance trail
Eleven skills take an agent from an idea to verified code through a PRD and RFCs kept as files in the repository
Comparison with similar tools
Entries that share the main search query, and how they differ.
README in short
The Chinese README presents the program as a layer that trains the agent to deliver projects reliably. It continues with installation, a five-minute start and the main features: experts, design system, pipeline, documents, checks, agent onboarding, knowledge base and audit.