Rubrik has launched Rubrik Agent Cloud for Anthropic’s Claude Code and Claude Cowork, geared toward organisations deploying Claude-based brokers in software program improvement workflows.
The service is designed so as to add oversight, restoration and coverage controls round autonomous brokers that may write, push and deploy code with out direct human motion. Rubrik positions the launch as a response to a safety hole between conventional enterprise controls and AI techniques that may act at machine pace.
The brand new service contains what Rubrik calls a governance engine for real-time management of autonomous brokers, alongside a list device supposed to point out threat, entry permissions and coverage violations throughout deployed brokers. It additionally features a function referred to as Agent Rewind, supposed to reverse unintended actions taken by customized brokers and brokers inbuilt instruments resembling Claude Code and Claude Cowork.
One other a part of the launch focuses on code repositories. Rubrik says it maintains immutable snapshots of GitHub and Azure DevOps repositories outdoors the repository setting, permitting organisations to revive a known-good state if model management can’t undo an undesirable change, resembling deleted branches or overwritten commit historical past.
The corporate can also be providing backup and restoration for the configuration that shapes how Claude brokers behave. That features system prompts, device permissions, abilities and key information used at organisation, repository and consumer ranges. The service displays for configuration drift and flags modifications that seem malicious or unauthorised earlier than they unfold.
The transfer displays a broader shift in software program improvement, the place AI brokers are taking over a bigger position in producing and modifying code. That has created new issues for safety and compliance groups as a result of present DevSecOps processes had been largely constructed round human approvals and guide assessment.
Rubrik recognized dangers together with rogue commits, repository ransomware, immediate injection and mental property exfiltration. In its view, these threats develop into more durable to include when an automatic agent could make modifications shortly and at scale.
Anneka Gupta, Chief Product Officer at Rubrik, commented on the demand the corporate is seeing round Anthropic’s fashions.
“Organisations are adopting Claude quicker than any agentic know-how now we have seen, and each safety chief asks the identical query: how will we keep in management when an agent can act?” stated Anneka Gupta, Chief Product Officer at Rubrik.
She stated the product is meant to present clients a approach to govern entry and get well from undesirable actions affecting each code and agent settings.
“Rubrik Agent Cloud provides organisations a resilience layer for Claude, which permits them to see what brokers can entry, govern what they do, rewind their actions, and get well each the code and the agent’s personal configuration when one thing is destroyed or tampered with. Working with Anthropic, a pacesetter in AI, lets us carry that management to clients from day one,” Gupta stated.
Safety focus
The launch underlines an rising marketplace for AI operational controls, with suppliers making an attempt to deal with governance, auditability and rollback for autonomous software program brokers. As companies experiment with AI instruments that may work together immediately with codebases and inside techniques, safety groups are below stress to trace what these techniques can entry and what they’ve modified.
Rubrik’s method combines monitoring with restoration. If an agent, or an attacker exploiting one, takes an motion outdoors the attain of regular model management, directors can restore repositories at both repository or organisation stage. Rubrik additionally says ransomware rollback is a part of the supply for code environments.
Anthropic hyperlink
The launch centres on Anthropic’s Claude instruments, that are more and more being utilized in coding and office assistant duties. Anthropic describes itself as an AI security and analysis firm and says the Claude household of fashions is used for work together with code understanding and safety evaluation.
For Rubrik, the tie-up expands its push into AI operations alongside its established work in cyber resilience and information safety. The corporate has been constructing merchandise geared toward serving to clients monitor, audit and get well AI-driven environments as using autonomous brokers strikes from experimentation to manufacturing.
The central challenge for patrons is more likely to be whether or not safeguards can preserve tempo with the rising autonomy of AI techniques. In software program improvement settings, meaning not solely stopping unauthorised or damaging modifications but in addition restoring techniques shortly when prevention fails.
Rubrik says its service tracks and restores the information and permissions that information agent behaviour, together with settings resembling CLAUDE.md, system prompts and gear entry. It additionally ties detected configuration drift to wholesome backup snapshots to assist orchestrated restoration.
That emphasis on each code and configuration displays rising recognition that AI brokers introduce a second layer of operational threat. The risk will not be restricted to what an agent modifications in a repository, but in addition extends as to whether the foundations governing that agent have themselves been altered with out approval.
Rubrik says Agent Rewind can reverse unintended actions throughout customized brokers and customary agentic improvement environments, together with Claude Code and Claude Cowork.