Charwork vs CodeConnect Remote

Charwork is a wireframe tool that draws interfaces out of plain Unicode characters. Pick a component from the palette, place it on the character grid, and build up a layout the way you would sketch one on graph paper. Everything you make is just text, so you can paste it into a README, a code comment, a design doc, a ticket, or a chat message and it renders anywhere a monospace font does. FEATURES - 23 component presets: buttons, inputs, selects, checkboxes, radios, toggles, tables, modals, browser frames, cards, navbars, tabs, progress bars, icons, images, dividers, alerts, breadcrumbs, avatars, lists, steppers, ratings, and skeletons - A 100 x 50 character canvas with click-to-place positioning - Hover preview, so you see where a component lands before committing to it - Undo and redo up to 50 steps - Export to a .txt file or copy the whole canvas to the clipboard - Works entirely offline; nothing you draw leaves your device Charwork does not collect any data, require an account, or show ads.

CodeConnect puts every coding-agent session running on your own computer into your hand. Works with Claude Code and OpenAI Codex. Steer an agent mid-run, answer the permission prompt it is blocked on, review its diff, or take over its terminal outright. Your sessions travel directly between the app and the CodeConnect daemon on your own machine. No server sits in between, and your code never leaves your own devices. The single CodeConnect-operated service is a content-free push relay, described in full in the privacy policy. WHAT IT DOES • A fleet view. Every agent, what it is asking for, and how long it has been waiting. • Approvals that show the real command. Risk-tiered: a read is one tap, while a destructive command needs a deliberate hold and a biometric confirmation. The command is never truncated or faded, because a hidden suffix is exactly where a dangerous argument hides. • Codex, with its own options. A Codex request is answered with the choices Codex actually offered, including the exact scope of "don't ask again". Stop a running turn, or send a message that starts a turn when Codex is idle and joins the running one when it is busy. • Nothing decides without you. A prompt is held open until you answer it. If the daemon cannot be reached, the decision falls back to your keyboard rather than being silently answered on your behalf. • Diff review that works on a small screen. Unified, monospace, word-level highlighting, and a comment back to the agent on any hunk. • Talk to it. The compose bar stages text into the agent's prompt; template chips insert rather than send. Dictation is handled by the system speech recognizer, and a transcript is always staged for your review, never sent automatically. • A real terminal. Attach to the live session over the connection the app is already paired on, and take over completely. • It never lies about state. Every fact shows its age, a stale link disables actions and says why, and "answered at the keyboard" is a state the app renders rather than a guess. WHAT YOU NEED • A computer running macOS 13 or later, with the free CodeConnect daemon installed from codeconnect.sh (0.7.0 or later for Codex). Apple silicon and Intel are both supported. • tmux on that machine. • A network path from your phone to it — the same Wi-Fi, or any private network you operate. Tailscale is a common choice, and is optional. • iOS 17 or later. CodeConnect is an independent tool and is not affiliated with, sponsored by, or endorsed by the vendors of the coding agents it supervises.