What happened
On Oct 8, Arcjet announced Arcjet Runtime Security for Coding Agents, extending its Agent Runtime Security platform to secure coding agents on developer endpoints. It gives security teams discovery of running agents, OPA/Rego policy enforcement via agent hooks (prompt-injection/PII/malware protection, blocking unapproved MCP/API servers, credential-exfiltration prevention) and auditable activity logs that pipe to existing SIEMs.
Why it matters
One of the first agent-agnostic runtime policy engines purpose-built for coding agents on dev endpoints — a mounting risk surface (agent-issued PRs, shell commands, MCP tools) that traditional firewalls and endpoint agents miss; positions runtime agent security as an OPA-driven policy layer rather than another endpoint agent.
Applicability
AppSec/platform teams running mixed coding agents (Claude Code, Codex, Cursor, Copilot) that want a single policy engine without standardising tools; roll out in minutes via provider-managed settings or MDM.