Pushary
Control panel that routes your AI coding agent's permission requests to your phone, Mac, or Slack for a tap-to-approve decision.
Overview
Pushary is a control panel that sits between your AI coding agents and the permission decisions they need a human for. When an agent like Claude Code, Codex, or Cursor hits a point where it needs sign-off (a delete, a deploy, a spend), Pushary pushes that decision to your phone, Mac notch app, or Slack as a tap-to-approve, tap-to-deny, or fill-in-a-value prompt, so you don't have to stay glued to a terminal to keep an agent moving. Permission policies let you auto-approve low-risk, reversible actions while still escalating anything destructive, and every decision lands in an audit trail you can review later.
It works with Claude Code, Codex, Cursor, Hermes Agent, Windsurf, Gemini CLI, and any MCP-compatible agent, plus Slack and n8n, and claims a sub-2-minute command-line setup. Pushary has been featured repeatedly in Product Hunt's AI daily digests (19-20 September 2026).
Key Features
Multi-Platform Delivery: Push notifications to iOS/Android, a Mac notch app, Slack, and a web dashboard.
Three Question Types: Confirm (yes/no), select (multiple choice), and free-text input prompts.
Tiered Permission Policies: Auto-approves safe, reversible actions while escalating risky ones (deletions, deployments, spending).
Audit Trail: A complete decision ledger of every approval and denial.
Fleet Management: Monitor multiple agents at once from one control panel.
Broad Integrations: Claude Code, Codex, Cursor, Hermes Agent, Windsurf, Gemini CLI, any MCP-compatible agent, plus Slack and n8n.
Pricing
Starting price
$9.99/mo
Two individual-developer plans, plus a separate B2B integration tier:
Agent: $9.99/month. 5,000 notifications/month. Preceded by a 3-day free trial (card required at signup).
Agent Pro: $19.99/month. Unlimited notifications, supports multiple simultaneous agents.
Partners (B2B, not the individual plan above): Launch $99/mo (5,000 end-users, 25,000 monthly requests, up to 3 team members); Growth $299/mo (25,000 end-users, 100,000 monthly requests, up to 10 team members); Scale $799/mo (100,000 end-users, 500,000 monthly requests, up to 25 team members); Enterprise custom pricing. All Partners tiers include a TypeScript SDK, REST API, MCP support, signed webhooks, and durable polling. First trial per workspace: 100 connected customers and 100 human requests for 3 days, card required.
Important: Partners is a paid B2B integration tier for companies embedding Pushary's approval layer into their own product. It is not an affiliate, referral, or commission program. No free tier exists at any level; every plan, including the trial, requires a card.
Pros
Broad agent coverage out of the box (Claude Code, Codex, Cursor, Hermes Agent, Windsurf, Gemini CLI, any MCP-compatible agent).
A genuinely tiered permission model, not just blanket approve/deny: auto-approves safe reversible actions and escalates risky ones.
Built-in audit trail for every decision, useful if you need to explain later what an agent was allowed to do.
Fast, CLI-based setup, claimed under 2 minutes on its own site.
A real B2B integration path via its Partners tier for companies wanting to embed the same approval layer into their own products.
Cons
No free tier; only a 3-day trial that requires a card upfront, then $9.99/mo minimum.
"Partners" is a paid B2B integration tier (roughly $99-799/mo) for companies embedding Pushary into their own product, not a referral or affiliate program, and no affiliate program of any kind is confirmed on the site.
Solves a narrow, specific problem (agent permission routing), useful only if you're already running agentic coding tools day to day.
Days to weeks old as a product; no independent reviews (G2, Trustpilot, etc.) to check its reliability or notification-delivery claims against.
The approval flow depends on you actually seeing and acting on the push promptly, a slow response still blocks the agent.
What Makes It Unique
Pushary's differentiator isn't that it sends notifications, several tools do that, it's the tiered permission-policy layer: auto-approving safe, reversible actions while escalating destructive ones (deletions, deployments, spending), combined with a full audit trail and fleet-wide monitoring across every major coding agent from one panel. That's a more complete human-in-the-loop control system than a simple approve/deny push alert.
Kay Score
6
/ 10
Tool Information
Pricing
$9.99/mo
Category
Coding & Development
Platform
Web / iOS / Android
Last Updated
