Anaconda Expands AI Platform With Agent Swarms and Autonomous Red Teams

Anaconda has expanded its AI platform with agent swarms, autonomous security testing and governed orchestration, combining technology from its Kilo Code, Enkrypt AI and Outerbounds acquisitions.

The new AI Workspaces support local development, multiple agents in VS Code, more than 500 models and over 19,000 vetted packages. The security layer adds autonomous red teaming across more than 300 attack categories, runtime controls that can approve, modify or block behavior, and an Agent Incident Registry.

Anaconda says 63 percent of AI-native builders it studied are moving toward agent swarms. Enkrypt AI separately found vulnerabilities affecting 73 percent of more than 25,000 MCP servers it scanned. Both figures are company research and should be evaluated with their methodologies in mind.

The strategic issue is governance across the full agent lifecycle. Development speed, dependency trust, tool permissions, runtime policy and incident evidence need to remain connected as agents move from experimentation into production.