autowright

agent
Security Audit
Pass
Health Pass
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Community trust — 15 GitHub stars
Code Pass
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Pass
  • Permissions — No dangerous permissions requested

No AI report is available for this listing yet.

SUMMARY

Make it easy to automate personal desktop tasks

README.md

Autowright logo Autowright

Make it easy to automate desktop tasks. You describe the job once and the app will run it forever.

Website: autowright.ai

Autowright is a desktop app for recurring personal automations. You describe the job in
plain words, a connected AI agent writes it as human-readable Python step scripts, you review and approve, and a
local scheduler runs those exact scripts on time. AI is used once, at authoring time.

Install

Download from autowright.ai or the
latest release:

  • macOS (Apple Silicon) - DMG, Developer-ID signed and notarized, or
    brew install --cask hansololz/tap/autowright
  • Windows (x64) - installer, unstable: early port, unsigned (expect a SmartScreen
    warning), and not yet tested on a wide range of machines
  • Linux (x86_64) - AppImage, unstable: early port, lags behind the macOS release,
    and known to fail on Ubuntu 24.04+ (AppArmor user-namespace restriction)

macOS is the supported platform today; the Windows and Linux builds are provided for early
testers and may break between releases.

Or build from source - see §18 in SPEC.md.

Features

  • Plain-words authoring and editing - the AI does the scripting, you do the deciding;
    nothing executes until you approve it.
  • Use your own agent - Claude Code, Gemini CLI, Codex, or OpenCode; OpenCode can drive
    a local Ollama model for fully offline drafting.
  • Real scheduling - cron with per-trigger timezones, one-shot triggers, run-on-app-start,
    and manual "Execute now". Runs even with the app closed (background service), with a
    missed-run policy for sleep and downtime.
  • Versioned automations - every approved edit is a new version; drafts run in isolation
    before you promote them.
  • Persistent memory with snapshots - automations keep state between runs, with automatic
    snapshots and one-click restore.
  • Live execution view - per-step status, streamed logs, and full execution history.
  • Menu-bar / tray surface - glance at what's running and fire jobs without opening the
    window.
  • Local and file-first - automations are YAML and Python on disk, secrets stay in the OS
    secret store (Keychain / Credential Manager), and everything runs on your machine.
  • Share automations - portable .autowright export/import, from the app or the CLI.
  • CLI with full app parity - author automations as files, execute and follow them,
    manage secrets and triggers. Headless- and agent-friendly with explicit per-automation
    secret grants.

Planned

  • Agent skill - drive Autowright straight from your agent chat session (create, edit,
    and run automations without opening the app); see skills/autowright/.
  • Headless pip package - run the backend and CLI without the desktop app.
  • GitHub sync - keep your automations in a repo and pull changes into the app.
  • Automation & agent marketplace - browse, share, and install automations and agents
    made by others.
  • More harness integrations and richer triggers - file-system changes, calendar
    events, and more.

Status

Early release, under active development. Feedback and issues are very welcome. SPEC.md is
the source of truth for the whole app; see §18 for the dev workflow.

Found a security problem? Please report it privately; see SECURITY.md.

Reviews (0)

No results found