Show HN: Agentctl – Terraform for your agent harnesses (github.com)

Agentctl, a new command-line tool, has been launched to streamline the management of AI agent harness configurations through a single declarative `workspace.toml` file. This tool operates in a way reminiscent of Terraform, offering commands like `validate`, `preview`, and `apply`, facilitating a seamless workflow across various AI tools, including OpenCode, Agy, Claude Code, and Codex. Its design allows users to define models, context, skills, guardrails, and harness-specific flags in a structured format, making it easier to maintain consistency across different environments. The significance of agentctl lies in its unified approach to managing configurations, which can greatly enhance efficiency and reduce errors in AI development workflows. Each harness acts as an isolated renderer module, simplifying the integration and modification of support for new models or tools. With features like environment variable expansion for secrets management, integrative skill handling, and atomic file writing, agentctl presents a robust solution for developers looking to enhance their AI projects. Built using the Rust ecosystem, this tool aims to aid in both single-user and CI environments, promising improved collaboration and productivity in the rapidly evolving AI/ML landscape.