日本語 
Keeper Security APAC Co., Ltd., a cybersecurity provider known for its zero-trust and zero-knowledge-based identity security and privileged access management ( PAM), has announced the launch of ” Agent Kit,” which enables secure automation of tasks in AI development environments.
Agent Kit is a suite of features that allows Keeper Secret Manager and Commander to directly integrate with leading AI coding agents such as Claude Code, Cursor, Codex, and GitHub Copilot. It enables secure automation of security settings and management tasks.
In recent years, while many companies have incorporated AI agents into their development processes, new challenges have also emerged. A particularly significant issue is the risk of privileged authentication information remaining in the AI’s prompt history. Traditionally, developers had to enter API keys and database authentication information via chat in order for AI agents to perform infrastructure operations. As a result, there was a risk that confidential information could be stored in external logs or training data.
Also Read: CrowdStrike unveils Charlotte AI AgentWorks ecosystem
Agent Kit addresses this challenge. The AI agent directly utilizes Keeper‘s CLI tools ( Commander and Keeper Secret Manager CLI ) and executes commands within the developer’s authenticated session. This allows sensitive information to be processed without being exposed externally.
Furthermore, for teams using hosted or integrated AI environments, we also offer Model Context Protocol (MCP) server integration (compatible with Docker and Node configurations). This mechanism allows you to retrieve secrets via a running MCP server without using the local CLI.
Even when an AI agent uses Keeper’s CLI tools, all operations are subject to role-based access control and audit logging, just as they would be when performed by a human.
SOURCE: PRTimes
English