Superagent

Superagent is a security product that now centres on protecting AI agents and the code they touch. Superagent's homepage, read directly, carries the tagline make your software secure itself and build your autonomous security factory. The docs describe four security workers: PR Security, Secure Dependency Updates, Red Team and Agent Guardrails. Targets are repositories, deployed web apps and agents, with infrastructure marked coming soon. Capabilities include Context Guardrails that score what agents consume before they trust it, Runtime Guardrails that monitor or block risky agent actions, a Security Factory, and an MCP install step. Customers named include Ubicloud, Evry Health, dotenvx, Firecrawl, Capchase, Nango and Mastra.

The docs pricing page returned a 404, and neither the homepage nor the docs overview mentions plans, prices, a free tier or a billing model, so I list pricing as contact sales. The open-source licence is not stated on the pages, and no company, funding or acquisition details appeared. The site shows no dates or changelog, so I could not confirm recent activity, though nothing says it is discontinued. Note that the project was previously known for open-source tools for building AI assistants safely, so older articles may describe a different product. Ask the vendor about pricing, data handling and which workers are generally available.

Superagent suits engineering teams shipping agents that need guardrails and red-team testing. Compare Lakera, LangSmith and Portkey. Browse the AI Agents & Automation tools directory for more.

Red teaming deployed agents

Adversarial tests.

Runtime guardrails

Monitor or block.

Superagent price?

Not published.

What does it do?

Secures agents and code with AI workers.

Open source?

Not stated on the pages.

Infrastructure scanning?

Coming soon.