What happened
On Oct 6 Anaconda expanded the Anaconda Platform (AI Dev Factory): Kilo Agent Swarms bring multi-agent development to VS Code, paired with Enkrypt AI's autonomous red-teaming agents (300+ attack categories adapting in real time), Enkrypt Runtime Guardrails that approve/modify/block risky agent and MCP behavior pre-execution, model risk scores shown in Kilo Code, an Agent Incident Registry of source-verified agent incidents, and an Advisory & SBOM API (beta) for agents/packages/MCPs. main-x (13,000+ vetted packages) is GA.
Why it matters
Anaconda is coupling agentic coding with continuous adversarial testing and runtime blocking in the builder's own IDE — a shift from scan-at-CI to in-loop security. Enkrypt's cited data (73% of 268k+ agent tools across 25k+ MCP servers vulnerable) shows how urgent agent/MCP supply-chain security is; this is one of the few comers putting runtime guardrails on MCP calls inside the coding tool itself.
Applicability
Enterprises standardizing AI Dev Factory workflows in Python/data-science shops should evaluate Kilo + Enkrypt guardrails to get continuous red-teaming and MCP runtime policy without a separate security console.