Pushary Documentation
Pushary sends your AI agent's questions and approvals to your phone. Set it up with one command, then approve, deny or answer from anywhere.
Pushary is the decision layer for AI agents. When an agent like Claude Code, Codex or Cursor needs a yes before a risky step, Pushary sends the question to your phone or your Mac. You tap Approve or Deny and the agent keeps working. If you build agents, your users can approve your agent's actions the same way.
npx pushary@latest setupRun it on the computer where your agent runs. It finds your agents, connects them, saves your API key and pairs your phone. The quickstart walks through each step.
Pick what you want to do
Use Pushary with my own agents
Claude Code, Codex, Cursor, Gemini CLI, Hermes and others ask you before risky steps. You answer on your phone or your Mac. One command sets it up.
Build approvals into my product
Your product's agent asks your users before a refund, an email or a delete. They tap Approve or Deny on their phone.
Most used pages
Supported agents
Every agent Pushary works with, and what it can enforce for each one.
Claude Code setup
Approve Claude Code's shell commands and edits from your phone.
Policies
Decide which actions run on their own and which ask you first.
REST API
Send a notification or ask a question with one HTTP call.
Pricing and limits
What each plan costs and the limits that apply.
Troubleshooting
Phone did not buzz? Agent did not ask? Start here.
Building a website instead?
If you run a website or store and want to send push notifications to your visitors, see the Push API for websites. It is a separate product on the same account.
See what changed recently in the changelog.