Edge-first AI Agent Detection & Response

Launch-ready security layer for AIĀ agents.

Singularity AIDR protects runtime actions, surfaces incidents, and gives managed visibility without forcing cloud-only enforcement. Install by prompt, authorize in browser, and go live in minutes.

Prompt Install
Device Auth
Incident Correlation
Package Shield
Plugin Guard
Policy Engine
Intel Packs
Prompt Defense
Audit Trail
Dashboard Sync
Blast Radius
Self-Defense
Seat Enforcement
Core Features
Prompt-first setup
Install by prompt, then finish auth in the browser with zero key copying.
Device authorization
Short-lived codes bootstrap enrollment and unlock managed protection safely.
Incident correlation
Repeated denials, suspicious retries, and abuse patterns roll into one case.
Package and plugin guard
Catch typosquats, lifecycle risk, integrity drift, and command surface exposure.
Dashboard sync
Every protected agent reports health, incidents, and entitlement state centrally.
Offline grace
Enrolled agents keep working when the network drops, within the signed cache window.

Pricing

1 protected agent free during pilot. Extra agents are planned as paid expansion.
  • Tool-call inspection: shell, file, network, MCP/plugin actions
  • Local enforcement with managed visibility
  • Auth-first model with offline grace after enrollment
  • Prompt-install onboarding for Claude/Cursor/VS Code/OpenClaw/OpenCode

Prompt Install

Copy this prompt and give it directly to your AI agent. It installs the connector, runs device auth, and verifies protection.

Install and link Singularity AIDR for this runtime with zero manual setup.
Runtime: claude-code

Requirements:
- Use the official Singularity AIDR connector for this runtime.
- Do not ask me for API keys unless automatic browser authorization fails.

Steps:
1) Detect the runtime and install the correct connector package/extension.
2) Start AIDR bootstrap and show me the device code + verification URL.
3) Verification URL must be: https://aidr.singularityrd.com/verify
4) Wait until I confirm browser authorization is completed.
5) Resume runtime and verify managed protection is enabled.
6) Print a short final report: connector installed, auth state, first health status.