Back to tools

cplt

cplt is an open-source kernel-level sandbox wrapper that helps developers and teams running local AI coding agents produce more controlled code changes and command outputs with lower secret-leak risk.

Tool categories
CodingDeveloper toolsAgent
Tool links

Tool overview

Based on the available evidence, cplt looks more like a project worth watching and testing in security-sensitive environments than a broadly proven default choice. What the evidence clearly supports is the official GitHub repo description: it wraps GitHub Copilot CLI, OpenCode, Google Gemini CLI, Pi, or a plain shell in a kernel-level sandbox so the agent stays within the project boundary and is blocked from some sensitive reads or risky actions. GitHub stars and forks indicate attention, not reliability, compatibility, or production readiness.

Its practical role is not to replace an IDE, a code completion model, or a full DevSecOps platform. A better analogy is a constrained execution layer around local or CLI-based coding agents. It may be mistaken for a generic permissions tool, but it is more specifically a safety wrapper for agent execution. That matters for teams worried about agents reading home-directory secrets, environment variables, or credential files, or accidentally pushing to main or merging changes while operating inside real repositories or controlled automation flows.

On cost and adoption, there is no official pricing evidence in the sources.

Related social content