AI workflow tools. Local-first. Reviewed in public.
Aglet is a single app that hosts small, focused tools — what we call aglets. Each aglet is a self-contained workflow, mostly authored by AI agents and reviewed by humans in public GitHub PRs before being added to the catalog. It starts with a few tools and grows in public.
Everything runs locally. No accounts, no telemetry, no tracking.
Open Aglet, get to work immediately. A focused starter set, no setup — more land in public.
Aglets are declarative IR + optional QuickJS scripts. No native code. No DOM. No fetch by default.
Every install shows declared permissions. Network access requires a URL allowlist visible to you.
Every aglet enters the catalog via GitHub PR with author identity, source diff, and reviewer signature.
Your data lives in a local SQLite database. No cloud, no analytics, no telemetry.
Download the app (macOS), or install the CLI via Homebrew:
# App — download the signed .dmg, drag to Applications:
# https://cdn.aglet.dev/dist/Aglet.dmg
# CLI (dev / authoring tool):
brew install agent-rt/tap/agletThe app self-updates via Sparkle. iOS / Android coming. Build from source: github.com/agent-rt/aglet.