Docs menu

Documentation / Get started

Install

Install co:op, keep it up to date, and verify a download yourself.

On this page

Overview

co:op runs on macOS and Linux, on amd64 or arm64, with Git and Docker:

Shell

curl -fsSL https://coop.dryga.com/install.sh | sh

The script puts coop in ~/.local/bin. It needs curl, tar, and sha256sum or shasum. If Docker is available, it also builds the box image and runs coop doctor. If not, install Docker and run those yourself:

Shell

coop build --egress open && coop doctor

To change what the script does, set these before sh, as in curl -fsSL https://coop.dryga.com/install.sh | COOP_NO_BUILD=1 sh:

Variable What it does
COOP_VERSION Install this release tag instead of the latest one.
COOP_BIN_DIR Install into this folder instead of ~/.local/bin.
COOP_NO_BUILD=1 Skip building the box image.

Container runtimes

Online providers need Docker, which supports the helper that keeps provider credentials outside the box. Apple container on macOS 26+ can run commands with open networking and no credentials. If both are installed, co:op uses Docker while it is running and Apple container otherwise; set COOP_RUNTIME=docker to always use Docker.

On a Linux host with SELinux, such as Fedora or RHEL, co:op starts its containers with SELinux labels turned off (label=disable), so a box can read your project without relabeling your files. Capabilities, seccomp, the box's user, mounts and filesystem protection still confine it, but SELinux no longer separates boxes from the host or from each other. A Compose service you start yourself keeps your compose file's own settings.

Build from source

Shell

git clone https://github.com/AndrewDryga/coop && cd coop && make install

Update

coop update updates the binary, then prepares the image for this project's network mode. Upgrading lists what to change when a new version needs it.

coop update and install.sh check a release the same way. Both always check the archive against the release's checksums.txt. When cosign is installed, both also verify the release signature and stop if it is missing or does not match. Without cosign, the signature is not checked; see Verifying a download to check it by hand.

Verifying a download

install.sh comes from the main branch, served by GitHub Pages at coop.dryga.com. You can also fetch it from https://raw.githubusercontent.com/AndrewDryga/coop/main/install.sh.

The script checks the archive against the release's checksums.txt with sha256sum or shasum, and stops if the file or the tool is missing. If cosign is installed, it also verifies the release signature, which proves checksums.txt came from the release workflow for that tag.

To verify by hand before you run anything, download the release files. Set VER to a tag from the releases page, and change darwin_arm64 for another platform:

Shell examples

VER=vX.Y.Z
ASSET="coop_${VER#v}_darwin_arm64.tar.gz"
base="https://github.com/AndrewDryga/coop/releases/download/$VER"
curl -fsSLO "$base/$ASSET"
curl -fsSLO "$base/checksums.txt"
curl -fsSLO "$base/checksums.txt.bundle"

# 1. checksums.txt is signed by the release workflow (keyless Sigstore):
cosign verify-blob checksums.txt \
  --bundle checksums.txt.bundle \
  --certificate-identity "https://github.com/AndrewDryga/coop/.github/workflows/release.yml@refs/tags/$VER" \
  --certificate-oidc-issuer https://token.actions.githubusercontent.com

# 2. the archive matches the now-trusted checksum (Linux: sha256sum -c -):
awk -v f="$ASSET" '$2==f{print $1"  "f}' checksums.txt | shasum -a 256 -c -

Uninstall

Quit editor sessions, loops and other coop commands first. Then, in each project:

  • Merge or save any fork you still need. coop fork ls lists them, and their folders sit next to the project in <project>-forks/.
  • If the project has services, run coop down. Add --delete-volumes to also delete their data, such as a database.

Remove co:op's containers, images and volumes from Docker. The volumes hold caches and any service data coop down kept, in volumes named coop-<project>-<id>_<name>. Each ls shows what the next command removes, so check it first:

Shell examples

docker ps -a --filter label=coop
docker rm -f $(docker ps -aq --filter label=coop)

docker image ls --filter reference='coop-*'
docker image rm $(docker image ls --filter reference='coop-*' --format '{{.Repository}}:{{.Tag}}')

docker volume ls --filter name=^coop-
docker volume rm $(docker volume ls -q --filter name=^coop-)

Removing an image by name keeps it if another tag still uses it.

Remove the binary and this machine's records: run history, network approvals and editor sessions. Use your COOP_BIN_DIR if you installed elsewhere, or command -v coop to find the binary.

Shell examples

rm -f ~/.local/bin/coop
rm -rf ~/.local/state/coop "${XDG_STATE_HOME:-$HOME/.local/state}/coop"

In each project you approved, delete the hidden approval file. Without the records above, a later install refuses to run there until it is gone:

Shell examples

rm -f .coop-network-approval

The last step deletes your settings and every account you signed in to with coop login, along with the API keys in agents/env, your presets and MCP servers. If the folder syncs through Dropbox or another service, it disappears from your other machines too. Copy anything you want to keep, then:

Shell examples

rm -rf "${COOP_HOME:-${XDG_CONFIG_HOME:-$HOME/.config}/coop}"

Your projects keep their .agent/ folder and the instruction files coop init added, which are yours to keep or delete. Also remove any coop completion line from your shell startup file and any editor setting that runs coop acp.

View this page on GitHub