Skip to content
AgentOpen-source agent

clippinator

Open-source GPT-4 multi-agent coding assistant that plans, writes and tests projects

Pricing
Open source
Platforms
CLI, Linux, macOS
GitHub stars
411
Last commit
3 years ago
Language
Python
Listing updated
today

Similar agents

About clippinator

clippinator is an open-source agent for coding. It is available on CLI, Linux, macOS, Windows, API and self-hosted.

The README says Clippinator is a Python project that develops code for or alongside a user. It is driven by a main agent called Taskmaster, which asks clarifying questions, has an Architect plan the file layout, and then hands implementation, debugging and testing to specialised subagents such as Writer, Frontender, Editor, QA and Devops.

Agents share a written architecture document, a project structure built with ctags, linter output and a persistent memory. They can write files, run bash and background processes, drive Selenium for browser testing, and fetch web pages. The author notes that it can make mistakes, so the intended way to work with it is to watch its output and give feedback with Ctrl-C while it runs.

Setup requires Poetry, an OpenAI API key and optionally a SerpAPI key for search. The README warns that long GPT-4 runs can be expensive in OpenAI API usage. It suits developers who want to hand a project to an autonomous agent and steer it at a high level.

What it does

Taskmaster orchestration
A main agent asks the user questions, then delegates work to specialised subagents.
Specialised subagents
Architect, Writer, Frontender, Editor, QA and Devops each have their own prompts and tools.
Architecture document
The Architect writes a file-by-file plan that all agents can read.
Linter-driven feedback
Linter output is fed to agents, and the linter command can be configured with SetCI.
Browser testing
Selenium tools let agents view pages, read console logs and interact with them.
Resumable sessions
Work stops and resumes from the last saved state, and feedback can be given with Ctrl-C.

Good for

  • Generating a small web service from a high-level description
  • Letting an agent scaffold and debug a Python project while reviewing its progress
  • Automating test runs for a web front end with Selenium
  • Experimenting with multi-agent coding workflows

Pricing

Open source

Free to run from source; OpenAI API usage is billed separately.

clippinator FAQ

Which model does it use?

The README says the agents are based on GPT-4 and require an OpenAI API key.

Can it search the web?

Optionally, yes. Adding a SerpAPI key lets the model use search.

How do I give it feedback while it runs?

Press ^C to send feedback to the main agent; pressing it during a subagent aborts that subagent.

Is clippinator free?

Yes. clippinator is open source, so you can use and self-host it for free.

What are the best alternatives to clippinator?

The closest alternatives to clippinator are gpt-pilot, Devon, munder-difflin, Docs and devika.

Where this comes from

  • Description: written from the project's README.