Argus by SoteroInc.See and control what your AI agents and MCP tools can reach. AI coding agents and MCP servers run with your credentials and your access. Argus puts a checkpoint in front of that: every agent gets a verified identity, every request is checked against policy you control, and every decision is recorded. Argus runs inside VS Code, alongside the agents it governs. What you can doRegister your agents. Give each agent or MCP tool a verified identity and a defined tool scope, so requests arrive attributable instead of anonymous. Classify your data sources. Tag each source — database, file store, API — with a sensitivity tier: Public, Internal, Confidential, or Restricted. Set policy. Decide which agents may reach which sources. Argus allows or blocks each request accordingly, before the agent's own code sees a result. Watch it happen. The activity view shows every request in real time — which agent, which resource, allowed or blocked, and why. Each request expands into a step-by-step timeline of what was checked and how long it took. Getting started
Requirements
Settings
CommandsAvailable from the Command Palette under Argus:
Privacy and securitySign-in happens in your system browser using PKCE. The extension never handles your password or your Google or Microsoft credentials. Tokens are stored in VS Code Support© 2026 Sotero, Inc. All rights reserved. Licensed under a proprietary license; see the LICENSE file included with this extension. |