First steps
You've installed Hydrate. The five things below are what most people do in the first ten minutes. Fill in the fields, then copy the generated script and run it in a terminal on the machine where Hydrate is installed. The page never runs anything — every line is yours to review and edit.
Anything you leave blank is skipped. The minimum useful run is
"project slug + repo directory" — that gives you a Hydrate
project with starter canon imported from your existing
CLAUDE.md / AGENTS.md.
1. License
If you have a license key, activate it now. Otherwise skip — Free tier works everywhere.
hydrate login2. Project
Tell Hydrate about the repo you'll be working in.
hydrate setupLLM-drafts a seed script (canon entries + backfilled sessions). Recommended.CLAUDE.md / AGENTS.md to summaries (kept as .pre-hydrate.bak)3. Import previous data (optional)
Coming from another machine or a teammate? Drop in a .hpack file path to seed your local store before anything else runs.
hydrate pack create emits these4. Git for team sync (optional)
Sharing canon with teammates? Bootstrap a git repo for the project's canon, push it once, and your team can pull on their machines.
hydrate-canon-<project>5. Verify
Run the 16-point health check at the end. Catches missing hooks, an unreachable daemon, license issues — anything that would cause a silent fail.
hydrate doctor at the endWhat the script does
- Sanity: verifies
hydrateis on PATH; aborts loudly if not. - License: activates with your key (if set), then optionally logs in to store your API key.
- Pack import: if you set a pack path, imports it before anything else so your project starts with that knowledge.
- Project init: creates the project from the supplied directory, optionally running
--dehydrateto shrink docs. - Team git: if you set a remote, bootstraps the canon repo, makes the first commit, pushes, and registers the path with Hydrate.
- Doctor: the 16-check health pass at the end. Anything red here is something to fix before moving on.