New updates have been reported about Arcjet.
Arcjet has expanded its AI runtime security platform with a new product, Arcjet Runtime Security for Coding Agents, positioning the company at the center of securing AI-assisted software development. The new capability extends Arcjet’s Agent Runtime Security from custom AI agents into popular coding assistants, enabling security teams to monitor agent activity on developer endpoints, enforce unified policies, and maintain detailed records of every action.
By integrating with tools such as Claude Code, OpenAI Codex, Cursor, Muse Code, and GitHub Copilot, Arcjet now provides an LLM- and agent-agnostic policy engine that operates consistently regardless of the coding agent or underlying model in use. The platform detects prompt injection, flags potential PII exposure, blocks risky API and MCP destinations, and can halt sessions showing signs of credential exfiltration, giving enterprises a centralized way to manage risk as developers adopt heterogeneous AI tools.
Arcjet embeds its controls directly into coding agent hooks and is powered by Open Policy Agent and Rego, allowing deterministic runtime enforcement and centralized policy updates without requiring organizations to standardize on a single AI provider. An optional endpoint agent can discover unmanaged or shadow coding agents, broadening coverage across developer machines and ensuring that all tools are brought under governance.
The company emphasizes that developer endpoints often have privileged access to sensitive code, data stores, and production-adjacent environments, making AI coding agents a critical new surface for security oversight. CEO and founder David Mytton said the rapid deployment of coding agents is outpacing traditional security capabilities, and Arcjet aims to provide “agent-native” controls to mitigate threats ranging from prompt injection to malicious MCP integrations.
Operationally, Arcjet can be deployed in minutes using existing AI provider management settings or device management platforms, lowering implementation friction for security and IT teams. Activity logs from Arcjet can be streamed into established SIEM systems, including Splunk, Datadog, SentinelOne, and Panther, allowing security operations to investigate, correlate, and audit coding agent behavior using their current monitoring and compliance workflows.
Strategically, this move deepens Arcjet’s role as a runtime security layer for AI applications by unifying visibility and control across both custom AI agents and developer-facing coding tools. Founded in 2023 and already deployed in more than 500 production applications, the company is backed by investors such as Plural, Andreessen Horowitz, Seedcamp, and several devtools and security-focused angels, positioning Arcjet to capitalize on the rapid enterprise adoption of AI coding assistants and the corresponding need for robust governance.
For executives, the launch signals a push by Arcjet to become a standard control plane for AI agent security across the software development lifecycle, with direct implications for risk management, compliance, and developer productivity. As organizations allow teams to choose their preferred coding agents rather than impose a single standardized tool, Arcjet’s unified policy framework and audit capabilities aim to reduce fragmentation, improve oversight, and provide verifiable evidence for regulators, customers, and internal stakeholders.
The expanded platform also addresses the governance gaps created by MCP servers, external APIs, and other integrations that coding agents can access, helping enterprises prevent data leakage and unauthorized operations at scale. By combining granular runtime controls with centralized policy management and SIEM integration, Arcjet is seeking to embed itself as a critical infrastructure layer for enterprises that are rapidly operationalizing AI-powered development workflows.
This product extension underscores Arcjet’s strategy of building a comprehensive runtime security stack for AI, spanning discovery, authorization, and evidence capture for every agent action, prompt, and tool call. As AI agents become more deeply integrated into both applications and developer tooling, Arcjet’s new offering is likely to influence security architecture decisions, procurement priorities, and AI governance frameworks for organizations investing heavily in AI-driven software delivery.
In the broader market context, Arcjet’s move highlights growing demand for specialized security solutions tailored to AI-driven automation rather than traditional human-centric workflows. If Arcjet can maintain its vendor-agnostic stance while scaling integration coverage and maintaining low-friction deployment, it could become a key enabler for enterprises seeking to balance rapid AI adoption with stringent security and compliance requirements.