Introduction
A terminal coding agent that takes permissions seriously
Most coding agents hold real power over your machine, gated by nothing
stronger than a system prompt. visvoai-cli is different: a full-screen
TUI agent (built on Textual) that reads,
edits, and runs code in your repo — with a permission model enforced by the
OS, not by politeness; delegatable subagents
with live logs; teachable skills;
MCP tool servers (Model Context Protocol); and
per-turn time-travel across both your files and the
conversation. Every one of those is a documented extension point — see
Extending VisvoAI CLI for the full menu.
It runs on the published visvoai-core
loop and visvoai-ai model layer,
unmodified — building your own agent product? This is the reference for how
far those two go.
Install
pip install visvoai-cli
export GEMINI_API_KEY=... # or ANTHROPIC_API_KEY / OPENAI_API_KEY / any compatible
visvoaiAll provider integrations ship in the box — the model picker
exposes a live catalog (Gemini, Claude, GPT, Together, Groq, OpenRouter, …);
you only supply a key for the provider you use.
visvoai "fix the failing test" runs a single-shot turn without the TUI —
see all commands and flags. For other install routes, see
Installation.
From source
uv tool install --editable path/to/visvoai-cli