CLI · TUI · local-first · MIT

TonAny

Terminal AI assistant that finishes the job in front of you — then hands off the rest.

Install See the TUI →

The terminal you actually get

Interactive TUI with tool cards, streaming bash, light memory, and a status footer — same shape as the app.

TonAny

TonAny · ~/acme (main) · anthropic · claude-sonnet

you Tidy this folder, fix the broken import in src/index.ts, and remember I prefer concise replies.

$ find . -maxdepth 2 -type f | head

./package.json
./src/index.ts
./src/utils.ts
./README.md

edit

src/index.ts

Fixed import path · 1 hunk

write

~/.tonany/agent/memory/profile.md

Preference saved via /skill:memory

tonany Done — folder grouped, import fixed, preference written to profile.md.

read / write / edit / bash — plus skills. Finish file and terminal work without leaving the TUI.

/skill:memory keeps preferences in local markdown under ~/.tonany/agent/memory — no vector DB.

/skill:handoff drafts a TonHandOff brief for TON (engineering loops) or TonWorker (office deliverables).

TypeScript extensions, skills, prompt templates, and themes — project-scoped under .tonany/.

Interactive TUI, print/JSON, RPC, and SDK embedding — same agent loop, different surfaces.

Anthropic, OpenAI, Gemini, DeepSeek, OpenRouter, Ollama-compatible — keys stay on your machine.

Short local work stays in TonAny. Specialists take the long jobs.

TonAny Universal assistant · whatever is in front of you right now

Node.js ≥ 22.19. Build from source, then run the TUI.

git clone https://github.com/toninfo/tonany.git && cd tonany && npm install --ignore-scripts && npm run hydrate:model-data && npm run build:offline

Sessions, settings, memory, and model keys live under ~/.tonany on your machine. Work context stays with you — less exposure, lower leak risk.