Blanket is a RESTy wrapper for long-running tasks. With a simple single-binary cross-platform install and embedded UI, API, CLI, and MCP interfaces, Blanket helps your hard-to-manage compute jobs to settle down so you can focus on composing those tasks into broader solutions.
Download a pre-built binary from GitHub Releases, or use one of the one-liner installers below.
Linux / macOS:
curl -sSfL https://raw.githubusercontent.com/turtlemonvh/blanket/master/scripts/install.sh | bashWindows (PowerShell):
irm https://raw.githubusercontent.com/turtlemonvh/blanket/master/scripts/install.ps1 | iexThe install script adds blanket to your PATH, sets up configuration,
loads default task types, and offers to set blanket up to run on boot
and to install AI-agent skills for drafting task types. See
installation for the environment variables that
control each of those, and where files land.
No sudo or administrator rights are needed for a default install
— everything goes under your home directory. Blanket is designed to be
easy to set up inside organisations with a semi-paranoid security
culture.
No internet access, or only local-user permissions? Each release
attaches blanket-bundle-<version>.tar.gz — binaries, checksums,
example task types and the install scripts in one file. See
offline install.
blanket upgrade --check # is a newer release available?
blanket upgrade --yes # install it and restart the server onto it
blanket rollback --yes # change your mindUpgrades are checksum-verified, take a database backup first, keep the
binary they replace so blanket rollback can put it back, and restart
the running server without losing the work in flight. See
upgrading.
# Start the server (uses the install-script-generated config by default)
blanketOpen the web UI at http://localhost:8773/.
# Submit a task — over REST or via the CLI
curl -s -X POST localhost:8773/task/ -d '{"type": "echo_task"}'
blanket submit -t echo_task
# Run a worker that accepts bash/unix tasks
blanket worker -t exec:bash,os:unixThat's it. For curl examples, file uploads, scripting, custom task types, and the full REST API, see the docs:
- Usage — submitting tasks, file uploads, the CLI, managing tasks at scale
- Task type definitions — TOML schema for authoring your own task types
- API reference — full list of REST endpoints
- Autostart — running blanket as a background
service that starts on login/boot, and
blanket uninstall - Task flow — task and worker state machines
- Upgrading —
blanket upgrade/blanket rollback, checksums and offline bundles, schema versions, backups,blanket backup/blanket migrate - MCP interface — expose blanket to MCP clients (agents), security considerations, and setup
See CONTRIBUTORS.md for development setup, build instructions, CI details, and code conventions.
MIT.