Honest comparison
Happy is a free, open-source mirror of Claude Code and Codex on your phone. Pushary is the cross-vendor control plane: set per-tool rules once, approve only what matters, and keep an exportable audit trail of every decision across every agent.
If you want a free, open-source, end-to-end-encrypted mirror of Claude Code or Codex on your phone, Happy is excellent. If you want author-once guardrails, an audit trail, and one tool across every agent you run, Pushary is the better fit.
Happy asks you to Allow or Deny each request, and a remembered choice lasts only for that session. Pushary lets you author the rules once: auto-approve safe reads, push on Bash, escalate git push, always deny rm -rf. The phone only buzzes for the decisions that actually matter, and the rule sticks across sessions and machines. That is the difference between babysitting two agents and trusting four.
Happy is end-to-end encrypted, which is great for privacy and also means the server cannot produce a queryable record of who approved what. Pushary keeps an immutable, attributed, exportable log of every question and every human decision, across every agent and machine. That is exactly what a team lead or a compliance owner needs, and it is the one thing an encrypted mirror structurally cannot give them.
Happy covers Claude Code and Codex. Pushary is one MCP server and one policy that also covers Cursor, Hermes, Windsurf, and any MCP client. One config, one queue, one audit trail, no matter which agent is blocked.
Happy is free, open source under the MIT license, and end-to-end encrypted, with a full real-time mirror of your session and voice. If you want an encrypted, open-source mirror of Claude Code or Codex on your phone and you do not need policy or audit, Happy is the better pick. Pushary makes the opposite bet: send the decision, not the transcript, and keep a record.
| Feature | Pushary | Happy |
|---|---|---|
| Per-tool permission policies (set rules once)POLICY | ||
| Immutable, exportable audit trailAUDIT | ||
| Works with Cursor, Hermes, Windsurf (any MCP client) | ||
| Works with Claude Code | ||
| Works with Codex | ||
| Approve from your phone | ||
| Kill switch to stop a runaway agent | ||
| Keeps your existing CLI (no wrapper to run) | ||
| Sends decisions only, not your code or transcript | ||
| Run multiple agents in parallel | ||
| Full real-time session mirror + voice | ||
| End-to-end encrypted | ||
| Open source | ||
| Price | $9.99/mo | Free (MIT) |
Happy (happy.engineering) is a free, open-source app under the MIT license that lets you control Claude Code and Codex from your phone or the web. It mirrors your session in real time with end-to-end encryption, supports parallel sessions, and adds voice. You run the happy CLI instead of claude or codex, and you bring your own Claude or Codex subscription.
Happy mirrors your whole session to your phone. Pushary routes only the decisions that matter, under per-tool rules you set once, and keeps an exportable audit trail of every question and human decision across every agent. Happy shows you everything. Pushary decides what reaches you and records every decision.
Happy is free and open source under the MIT license. Pushary's Agent plan is $9.99/mo with a 7-day free trial. The paid line is what funds the policy engine, the immutable audit trail, and cross-vendor support beyond Claude Code and Codex.
Happy is end-to-end encrypted, so its relay only ever sees encrypted data. That is a strong privacy story, but it also means there is no server-side audit trail to query or export. Pushary is not end-to-end encrypted by design: it sends only the decision (the question text and the tool name), never your code or your transcript, and keeps that as a queryable, exportable record. Different trade-off: Happy maximizes encryption, Pushary keeps an auditable record while minimizing what leaves your machine.
Yes. They are independent. You could run Happy for a full end-to-end-encrypted session mirror and Pushary for per-tool policies, cross-vendor approvals, the kill switch, and the audit trail.
Set your guardrails once, approve from your phone, and keep an exportable audit trail. Works with Claude Code, Codex, Cursor, Hermes, and any MCP client. 7-day free trial.
