REPLbot is a bot for Slack and Discord that allows you to control a REPL or shell from a chat. It comes with a few REPLs (Go 🥳, Java, NodeJS, PHP, Python, Ruby, Scala, Kotlin, C++, Ubuntu, Debian, Alpine, Arch Linux, Amazon Linux, CentOS & Fedora), but you can easily make or bring your own.

I thought it might be a fun way to collaboratively dabble with a REPL in a team. Yes, I could have gone for a terminal in a browser, but there's nothing like having it right there in Slack. Mainly I did it because it was fun though. 😄

I use tmux and the tmux capture-pane command to run most of the show. It's simple, but effective. In the first iteration I tried using a pseudo terminal (pty) directly, but with all the escape sequences and commands, it was getting kinda tiresome, and I was spending time with stuff that I didn't want to spend time with (though I learned a lot!). And tmux does its job so well.

The actual REPLs are just simple scripts (see script.d folder), so they could be anything you like. I highly recommend using Docker to provide somewhat of an isolation, though you'll probably still need to trust the people using the bot if you give them an entire REPL.

Scala, split mode (Discord)	Ubuntu running `htop` (Discord)	C++, channel mode (Discord)
Blinking cursors, choo choo 🚂 (Slack)	Custom shells, multiple sessions (Slack)

After installing REPLbot, you may use it by tagging @replbot in Slack/Discord. For the most part it should be pretty self-explanatory:

To start a session with the default settings, simply say @replbot java to start a Java REPL. There are a few advanced arguments you can use when starting a session.

REPLbot can run more or less arbitrary scripts and interact with them -- they don't really have to be REPLs. Any interactive script is perfectly fine, whether it's a REPL or a Shell or even a game. By default, REPLbot ships with a few REPLs. To extend the REPLs you can run, simple add more scripts in the script-dir folder (see config.yml).

Here's a super simple example script:

#!/bin/sh
# Scripts are executed as "./script run <id>" to start the REPL, and as "./script kill <id>"
# to stop it. Not all scripts need the "kill" behavior, if they exit properly on SIGTERM.
case "$1" in
  run)
    while true; do
      echo -n "Enter name: "
      read name
      echo "Hello $name!"
    done 
    ;;
  *) ;;
esac

In all likelihood, you'll want more isolation by running REPLs as Docker containers. Here's the PHP REPL script that REPLbot ships with (not shortened):

#!/bin/sh
# REPLbot script to run a PHP REPL.
#
# Scripts are executed as "./script run <id>" to start the REPL,
# and as "./script kill <id>" to stop it.

DIR="$(cd -- "$(dirname "$0")" >/dev/null 2>&1 && pwd -P)"
case "$1" in
  run) docker run --rm --name "$2" -it php ;;
  kill) "$DIR"/helpers/docker-kill "$2" ;;
  *) echo "Syntax: $0 (run|kill) ID"; exit 1 ;;
esac

When a session is started, you can get a list of available commands by typing !help (or !h). To exit a session at any point in time, type !exit (or !q).

Sessions can be recorded using asciinema, and can even be automatically uploaded to either asciinema.org or your private asciinema-server (see install instructions).

After a session exits, the recording is then attached to chat as a ZIP file and optionally as a link to the asciinema server. This behavior can be controlled by the default-record and upload-recording option in the config.yml file, as well as the record/norecord option when starting a session.

Here's the asciinema.org link of the example I recorded, as well as the ZIP archive with the recording.

Entering commands via Slack or Discord can be quite cumbersome, so REPLbot provides a web-based terminal (powered by the amazingly awesome ttyd). If enabled, a unique link is created for each session, which provides a read-only or writable web terminal.

This behavior can be controlled by the default-web option in the config.yml file, as well as the web/noweb option when starting a session. While a session is running, it can be toggled using the !web command.

You can share your local terminal window in Slack or Discord using the share feature. It's quite cool, although it's really got nothing to do with REPLs 🤷. It also has to be specifically configured in the config.yml file using the share-host option, since it needs direct communication between the client and REPLbot.

You can specify if you want the session to be started in the main channel (channel), in a thread (thread), or in split mode (split) using both channel and thread. Split mode is the default because it is the cleanest to use: it'll use a thread for command input and the main channel to display the terminal.

You can set the terminal window size when you start a session by using the keywords tiny (60x15), small (80x24), medium (100x30), and large (120x38). The default is small. You may also resize the terminal while the session is running using the !resize command.

When starting a session, you can choose whether to trim empty space from the terminal session (trim), or whether to show the entire terminal window as it would appear in a terminal emulator (full). The default is full, as trim mode can get awkward when the terminal is expanded and the collapsed again.

Please check out the releases page for binaries and deb/rpm packages.

Requirements:

• A modern-ish Linux, preferably Ubuntu 18.04+, since that's what I develop on -- though it also runs on other distros.
• tmux >= 2.6 is required, which is part of Ubuntu 18.04 (but surprisingly not part of Amazon Linux!)
• docker for almost all scripts REPLbot ships with
• asciinema if you'd like to record sessions
• ttyd if you'd like to use the web terminal feature

Creating a REPLbot Slack app:
REPLbot requires a Slack "Classic App (bot)", because of its use of the real time messaging (RTM) API. To create a classic app and acquire a Slack bot token, follow these steps:

1. Create a classic Slack app
2. In the "App Home" section, add a "Legacy bot user"
3. In the "OAuth & Permissions" section, click "Install to Workspace" (this may require workspace admin approval)
4. Copy the "Bot User OAuth Token" starting with "xoxb-..."

Creating a REPLbot Discord app:

1. Create a Discord app
2. In the "Bot" section, click "Add Bot" and disable "Public Bot"
3. In the "OAuth2" section, click "Add Redirect" and type a URL (even https://google.com is fine), select the scopes "bot" and "messages.read", and the permissions "public threads", "private thread", "send messages", "manage messages", "manage threads". Click "Save changes".
4. Copy the OAuth2 URL and navigate to it in the browser and authorize the app.
5. In the "Bot" section, copy the token and paste it here

Installing replbot:

1. Make sure tmux and probably also docker are installed. Then install REPLbot using any of the methods below.
2. Then edit /etc/replbot/config.yml to add Slack or Discord bot token. REPLbot will figure out which one is which based on the format.
3. Review the scripts in /etc/replbot/script.d, and make sure that you have Docker installed if you'd like to use them.
4. If you're running REPLbot as non-root user (such as when you install the deb/rpm), be sure to add the replbot user to the docker group: sudo usermod -G docker -a replbot.
5. Then just run it with replbot (or systemctl start replbot when using the deb/rpm).

Debian/Ubuntu (from a repository):

curl -sSL https://archive.heckel.io/apt/pubkey.txt | sudo apt-key add -
sudo apt install apt-transport-https
sudo sh -c "echo 'deb [arch=amd64] https://archive.heckel.io/apt debian main' > /etc/apt/sources.list.d/archive.heckel.io.list"  
sudo apt update
sudo apt install replbot asciinema

Debian/Ubuntu (manual install):

sudo apt install tmux
wget https://github.com/binwiederhier/replbot/releases/download/v0.6.4/replbot_0.6.4_amd64.deb
dpkg -i replbot_0.6.4_amd64.deb

Fedora/RHEL/CentOS:

# Make sure that "tmux" is installed
rpm -ivh https://github.com/binwiederhier/replbot/releases/download/v0.6.4/replbot_0.6.4_amd64.rpm

Docker:
You can configure the Docker image by mounting a config directory (containing config.yml) to /etc/replbot. To be able to use most of the pre-packaged script.d REPLs (to be mounted to /etc/replbot/script.d), you'll need to give the REPLbot image access to the Docker socket file. This allows the container to spin up other containers on the host. This is a security risk and considered bad practice, but it's the only way.

# Config and scripts live on the host
wget https://github.com/binwiederhier/replbot/archive/refs/tags/v0.6.4.tar.gz
tar zxvf v0.6.4.tar.gz
sudo mkdir /etc/replbot
sudo cp -a replbot-0.6.4/config/{script.d,config.yml} /etc/replbot
vi /etc/replbot/config.yml
  # Configure at least "bot-token"
  # To support web terminal, set "web-host" (e.g. to localhost:31001)
  # To support sharing, set "share-host" (e.g. to localhost:31002)
  
# Then you can start the REPLbot docker image (assuming web-host listening
# at 31001 and share-host at 31002)
docker run --rm -it \
  -v /etc/replbot:/etc/replbot \
  -v ~/.asciinema:/root/.config/asciinema \
  -v /var/run/docker.sock:/var/run/docker.sock \
  -p 31001:31001 \
  -p 31002:31002 \
  binwiederhier/replbot

Please note that if you'd like to associate the asciinema videos to an account, you need to set up and map the config directory properly. For me, that worked by installing it on the host, running asciinema rec once and then associating the install-id located in ~/.asciinema/install-id. Inside the docker image, asciinema expects the config directory to be in ~/.config/asciinema.

Go:

# Be sure "tmux" is installed
go get -u heckel.io/replbot

Manual install (any x86_64-based Linux):

# Make sure that "tmux" is installed
wget https://github.com/binwiederhier/replbot/releases/download/v0.6.4/replbot_0.6.4_linux_x86_64.tar.gz
sudo tar -C /usr/bin -zxf replbot_0.6.4_linux_x86_64.tar.gz replbot

Building replbot is simple. Here's how you do it:

make build-simple
# Builds to dist/replbot_linux_amd64/replbot

To build releases, I use GoReleaser. If you have that installed, you can run make build or make build-snapshot.

I welcome any and all contributions. Just create a PR or an issue, or talk to me on Slack.

Made with ❤️ by Philipp C. Heckel, distributed under the Apache License 2.0.

Third party tools/libraries:

• tmux (ISC) is used to run most of the show
• docker is used for almost all scripts REPLbot ships with
• asciinema (GPLv3) is used to record sessions
• ttyd (MIT) is used to run the web terminal
• github.com/urfave/cli/v2 (MIT) is used to drive the CLI
• discord-go (BSD-3) is used for the Discord communication, specifically these two these two pull requests for thread support
• slack-go (BSD-2) is used for the Slack communication
• gliderlabs/ssh (BSD-3) is used for remote port forwarding for terminal sharing
• stretchr/testify (MIT) is used in most of the tests

Code and posts that helped:

• tmux sharing across hosts, thanks to QuAzI