Unterm is a terminal that AI agents can drive, plus the cockpit to run them from. It exposes 151 MCP methods, an agent Inbox, git-worktree fleets, diff review and rollback, and one-click updates. It runs on macOS, Linux and Windows, is local-first and free, and is released under MIT. Claude Code, Codex, Gemini CLI, Cursor and Aider can all control it over MCP.
Be the first to rate Unterm. Your feedback helps other users decide.
Run agents in separate git worktrees and let the Inbox surface the one that needs you.
Let any MCP-capable agent control the terminal over 151 methods.
Inspect diffs from agent runs and roll back the ones you do not want.
Similar tools in the same category — compare side-by-side
Mnestiq is a flight recorder for AI agents: signed, tamper-evident records of what an agent saw, dec...
QAmcp tests any MCP server in one command, covering tools, safety labels, hidden instructions, bad i...
mcpspan is open-source, self-hosted analytics for MCP servers. You add one line to your MCP server a...
Guides and comparisons about code tools
Nine new tools aimed at the same wall from different angles: indexed wikis, stale memory, knowledge graphs, reversible compression and multi-agent rooms.
Eight new tools that decide where a coding agent runs, how many run at once, and how you keep working in the same file it is editing.
Eight new tools for what sits under an agent: deterministic memory, an agentic database, distributed inference, hardware checks and spend guards.
Nine new tools that put something deterministic between an agent and what it touches: gateways, policy firewalls, AST breakers, SQL guards and usage ceilings.
Nine new tools for evidence rather than dashboards: signed records, decision logs, MCP analytics, protocol tests, sandboxes and long-running agent supervision.
