How to help

Help with issues, copy review, translations, model testing, and pull requests.

Pepe is young. Small, focused help already matters: opening a clear issue, answering questions, reviewing copy, testing providers, improving translations, or sending a PR.

Useful ways to help

Open good issues. Say what you tried, what you expected, what happened, and include relevant commands or logs.
Answer issues. Reproduce bugs, ask for missing details, and confirm whether a fix works.
Improve copy. Cut long explanations, replace literal translations with natural wording, and fix examples that feel artificial.
Translate. Keep English, Spanish, pt-BR, and pt-PT aligned. Translate reader text; preserve commands, tool names, payloads, and APIs.
Test models. Confirm streaming and tool calling on OpenAI, OpenRouter, Groq, DeepSeek, Together, Mistral, Ollama, LM Studio, vLLM, and other providers.
Send small PRs. One bug, one page, one translation, or one improvement per PR is easier to review.

From fork to PR

Fork the repository on GitHub.
Clone your fork:

git clone git@github.com:YOUR_USER/pepe.git
cd pepe

Add the original repository as upstream:

git remote add upstream https://github.com/pepe-agent/pepe.git
git fetch upstream

Create a branch from master:

git checkout -b docs-improve-quickstart upstream/master

Install dependencies and run tests:

mix deps.get
mix test

If you are changing the website:

cd website
npm install
npm run dev

Make the change. For docs and copy, review the other languages when the same page exists in them.
Run the final check from the project root:

mix precommit

Commit and push to your fork:

git add .
git commit -m "Improve quickstart copy"
git push origin docs-improve-quickstart

Open a pull request against pepe-agent/pepe:master. Explain what changed, why it changed, and link the issue if there is one.

Good PRs

A good PR is small, scoped, and easy to review. For code, include a test when behavior changes. For docs, prefer short sentences, real examples, and links to deeper pages instead of repeating everything.

Help with models

Provider reports are especially useful. The ideal report includes:

provider and model tested;
command used to configure it;
output of pepe model test;
a simple prompt that streamed a response;
a prompt that required a tool, such as reading a file or searching the web.

If something fails, open an issue with that context. Even a “tested and works” report helps track which integrations are healthy.

← Evals