Distribution
Share a reflex as a local file, by URL or GitHub, or through the commons — all installed with arx add.
A reflex is a single file, so sharing it is easy. arx add resolves four kinds of source and writes the result into .reflex/.
The commons (by name)
Official reflexes are installable by short name from the commons:
arx add no-force-pushadd resolves the name against the catalog at agentreflex.dev/registry, downloads the bundle, and wires it in.
GitHub or a URL
Point add at a file in a repo or any URL — great for sharing your own before (or instead of) the commons:
arx add github:you/reflexes/no-secrets.mjs
arx add https://example.com/reflexes/no-secrets.mjsgithub:owner/repo/path fetches that file from the repo's default branch.
A local file
Reference a file you already have, or one elsewhere on disk:
arx add ./no-force-push.mjs # already in .reflex/
arx add ../shared/no-secrets.mjs # copied in from elsewhereHow it lands
Every add writes the reflex into .reflex/ and records it in .reflex/config.json. It's a plain file in your repo from then on — read it, commit it, and your whole team gets it on clone (arx install re-wires the agent hooks).
A reflex is code that runs with your permissions. Pull official reflexes by name, or read what you add from a URL first — see the trust model.
Publishing to the commons
To make a reflex installable by short name, contribute it to the official catalog — see Contributing.