Install

Install Pepe and run the guided setup before creating agents.

Install the pepe binary, then run the guided setup. It creates the config file, connects a model, and creates your first agent.

1. Install

One command installs the pepe binary.

curl -fsSL https://pepe-agent.com/install.sh | sh

Check it landed:

pepe help

Pepe stores its setup in ~/.pepe/config.json. There is no database to run.

2. Guided setup (the fast path)

pepe setup walks you through everything: signing in to your model provider, picking a model, creating your first agent, and optionally connecting a channel.

pepe setup

If you prefer manual steps, use the model, agent, and channel pages. Both paths write the same config.

Secrets stay out of the file. When Pepe asks for an API key it accepts a ${ENV_VAR} reference, for example ${OPENROUTER_API_KEY}. The reference is what gets written to ~/.pepe/config.json. The real value is read from your environment at run time and is never stored expanded.

Docker

Prefer a container? docker pull ghcr.io/pepe-agent/pepe (amd64 and arm64). It needs a volume and a dashboard password. Both are covered, along with how to give the agent extra tools inside the container, on the Docker page.

Uninstall

Remove the binary; add the config directory to also drop every model, agent and credential you set up.

rm ~/.local/bin/pepe
rm -rf ~/.pepe   # optional - also drops your config

(~/.local/bin is the default install dir; it’s wherever $PEPE_BIN_DIR pointed to if you overrode it.)

← Introduction Docker →