llmmanorg/homebrew-tap

Homebrew tap for llmman — run any agent on any model, models stored as OCI images.

Install

brew install llmmanorg/tap/llmman

The fully-qualified name does the brew tap for you and trusts just this formula (Homebrew does not load formulae from third-party taps otherwise).

Supported platforms (the platforms llmman publishes builds for):

Platform	Architecture
macOS	Apple Silicon (arm64)
Linux	x86_64
Linux	aarch64

Intel macOS is not supported — llmman publishes no x86_64-apple-darwin build. cargo install llmman builds from source there (needs Rust and Go).

Run as a service

The CLI starts the daemon on demand; to keep it running across logins:

brew services start llmman

This runs llmman serve under launchd or systemd, restarts it if it exits, and logs to $(brew --prefix)/var/log/llmman.log. It listens on http://127.0.0.1:17434; to change that, put LLMMAN_HOST=... (plus LLMMAN_API_KEYS=... for a non-loopback bind) in ~/.homebrew/services/llmman.env ($XDG_CONFIG_HOME/homebrew/... if set), restart the service, and export the same variables in your shell so the CLI talks to it rather than starting its own daemon.

Run brew services restart llmman after brew upgrade llmman. Homebrew does not restart services on upgrade, and the CLI replaces a daemon whose version differs from its own, which leaves the service unable to bind until it is restarted.

Versions

Every commit that passes CI on llmman's main is a release, versioned MAJOR.MINOR.<commit count> (e.g. 0.1.324), and this formula is updated to it within minutes. brew upgrade llmman therefore tracks main; there is no separate stable channel.

This formula is generated

Formula/llmman.rb is rendered by packaging/render.sh in the main repo and pushed here by its CI on every release. Edits made directly in this repo are overwritten by the next release — change packaging/homebrew/Formula/llmman.rb.in upstream instead.