Remote approvals
Send an approval request to your phone, Slack, or the web app when an agent reaches a step that needs your decision, then continue the task after you respond.
Pushary
ClaimPushary is a human-in-the-loop control panel for AI agents that sends approval requests to your phone, Slack, or the web app so work can continue when permission is needed.
AI agent approval and control panel
Pushary is a control panel for AI agents that routes permission requests and task-completion alerts to your phone, Slack, the web app, and an audit trail. It is built for human-in-the-loop workflows where an agent needs a quick yes, no, or other response before it can continue.
The product is aimed at people running agents in tools such as Claude Code, Codex, Cursor, Hermes, Windsurf, Gemini CLI, and other MCP-compatible clients. Its main job is to remove the need to watch a terminal for prompts, so long-running work can keep moving while you are away.
Core capabilities
Flexible question types
Support yes/no confirmations, multiple-choice questions, and free-text replies so agents can ask for the right kind of input at the point of uncertainty.
Policy controls
Apply per-agent permission policies such as auto-approve for safe actions, escalation for risky actions, and fallback behavior when you do not answer in time.
Audit trail and session tracking
Track what each agent asked, which response was given, and what happened next in a full audit trail or session record.
Broad agent support
Use the same approval workflow from Claude Code, Codex, Cursor, Hermes, Windsurf, Gemini CLI, and any MCP agent, with setup that detects supported agents automatically.
Workflow integrations
Run the same approval pattern inside Slack or n8n workflows, or embed approvals into your own product through the Partner offering.
Practical use cases
Approve risky agent actions remotely
A developer kicks off a refactor, deploy, or shell-command-heavy task in Claude Code or Codex and wants approval prompts sent to a phone instead of waiting at the terminal.
Manage a multi-agent workflow
A person running several agents wants to see which session is waiting, manage multiple concurrent requests, and keep a record of who approved what.
Move approvals into team tools
A workflow owner wants approvals to happen in Slack or n8n instead of only in a local terminal, so the same decision can be answered from the collaboration tool already in use.
Embed approvals into a product
A product team wants to ask end users for approval inside its own AI product, using Pushary as the approval layer rather than building the human-checkpoint flow from scratch.
Keep long-running jobs moving
Someone steps away from a long task and wants to come back to a finished result, with the agent automatically handling safe steps and only escalating the decisions that matter.
Pros and Cons
Pros
- Lets an agent pause for permission and resume after a tap, which fits long-running tasks that would otherwise sit idle.
- Supports several response modes, including approve, deny, multiple choice, and text input.
- Includes policy controls for safe actions, risky actions, and timeouts when you are unavailable.
- Provides an audit trail or session history so approvals are recorded instead of buried in chat logs.
- Offers multiple delivery surfaces, including phone, Slack, the web app, and workflows such as n8n.
Cons
- The strongest setup and workflow details in the source are centered on CLI-based agent integrations; broader platform coverage is mentioned more generally than it is documented in detail.
- Some plan-specific limits and team-record distinctions are visible in pricing, but the source does not fully document every retention, notification, or history constraint across all surfaces.
- The native mobile app is mentioned as being in internal testing, so the phone workflow appears to rely on notifications and web-based access in the documented flow.
FAQ
Can I approve an AI agent from my phone?
Pushary routes the approval question to your phone. When an AI agent pauses for permission, you can approve, deny, choose from multiple options, or enter free text from the lock screen or phone notification.
How do I connect an AI agent to Pushary?
The docs describe two ways to connect agents: a setup wizard that configures MCP and permission hooks automatically, or manual configuration for supported agents such as Claude Code. For unsupported clients, Pushary also documents MCP and REST options.
Does Pushary have a trial or paid plans?
The pricing page shows a 7-day free trial on all plans, with paid plans available afterward. The main consumer plans start at $9.99 per month, and there is also a Team plan and a Partner plan for embedding approvals in products.
What kinds of questions can Pushary handle?
Pushary supports question types such as yes/no confirmation, multiple choice, and free-text input. The Claude Code guide also describes rich-context notifications when tasks complete or fail.
Does Pushary send my code to its servers?
The source states that Pushary sends decisions, not transcripts, and that only the question itself is sent to its servers. It also says missed questions can be handled by configured rules and that questions expire after 10 minutes.
Quick Facts
- Category
- Developer Tool / Productivity
- Primary use
- Human-in-the-loop approvals for AI agents
- Supported surfaces
- Phone, Slack, web app, audit trail
- Known agents
- Claude Code, Codex, Cursor, Hermes, Windsurf, Gemini CLI, MCP agents
- Pricing
- Paid plans start at $9.99/month after a 7-day free trial
- Setup
- One-command CLI setup is documented for supported agents
Pushary Alternatives
AI Magicx
AI Magicx is a unified AI workspace for chat, image, video, voice, music, email and developer tasks, helping teams and creators manage multiple models in one place.
Orca
Orca is an Agent Development Environment for shipping with coding agents, running multiple CLI agents in parallel across isolated worktrees, with desktop and mobile workflows.
Paper
Paper is a design tool that connects canvas, code, and AI agents so teams can create, share, and ship work in one workflow. Includes desktop app and MCP access.
CREAO
CREAO is an AI agent platform for turning recurring business work into reusable workflows, with schedules, approvals, and tool integrations.
blop
blop is a QA agent that writes browser tests as code in your repo, runs them in CI, clusters repeated failures, and can open PRs to fix broken tests.
RLAMA
RLAMA is a local AI platform for building RAG systems and intelligent agents on macOS, Linux, and Windows, with HTTP API support.