Show HN: AgentTerm expands AI CLI sessions into visual workspaces

Summary

AgentTerm is an MIT-licensed terminal built around AI command-line interfaces. It gives each AI CLI session its own rich visual workspace while preserving the familiar terminal and existing CLI functionality. The author describes using it for planning, code review, and session continuity across projects. During planning, a user can open the plan in the terminal, state their intent, and have the agent respond and edit the plan in place. Before pull requests, a curated review presents changes that require human judgment, explains the agent’s reasoning, and includes relevant unchanged code for context; the user can comment on either the code or explanation and ask the agent to make and explain fixes. AgentTerm can also recognize and resume sessions, including continuing work from a phone. Its visual features appear when needed and recede afterward, and the project is intended to be extended for individual or team workflows. The repository provides demos and setup instructions, including separate configuration for phone access and IDE integration. It has no telemetry or account requirement, supports macOS and Windows through WSL, and is released under the MIT license.