Use with your assistant

Add your board to your assistant app once, and it can put things on your screen whenever you ask.

Board addresshttps://botcanvas.us/connect

Hands-free in your Tesla: Say “Hey Grok, open BotCanvas and pair it.” That one command handles everything: you don’t ask for a pairing code or type one. Grok opens the car’s browser with your board already paired and a session ready, without you touching the screen. Then keep talking: ask Grok to show things on the board while you drive.

One click:

Add to ClaudeAdd to CursorAdd to VS Code

Or paste these instructions into your assistant (the same rules the board address gives it):

Use the BotCanvas connector tools to show things on my screen (address https://botcanvas.us/connect, notes at https://botcanvas.us/llms.txt). Don't write code or curl.

ONE PAGE RULE (most important): answer each request with ONE full-screen push_html page with pictures (a hero picture plus cards, or list-thumbs), not several separate push_text, push_link or push_image items. Only push separate items when I explicitly ask for them.

IMAGE URL RULE: every picture link must be a clean, stable, direct file link. Avoid apostrophes, spaces, parentheses or other special characters in the path; prefer names with only letters, numbers, hyphens and underscores, or use Wikimedia Special:FilePath with full percent-encoding (an apostrophe is %27). If a link looks problematic, swap it for a clean alternative. Before pushing, check that each picture link returns HTTP 200 with an image content type. If the push result lists missing pictures, push the page again with replacements and delete the old one.

The board is a glance screen, not a chat window. Explanations go in the chat or out loud. The board must only get big, picture-first content that I can take in at a glance. Don't copy your chat answer onto the board.

When I say "show it on BotCanvas", "show it", "put it on the board", "open the canvas", "on the canvas", "on my screen" or "on the car screen", I mean BotCanvas (not a chat canvas feature). If the board isn't paired, call get_pin. Only if you control the browser on my screen (like the browser in my Tesla), open https://botcanvas.us/?code=<code> there. Never open that link in your own browser or tool: that uses up the code. Otherwise tell me the code to type at botcanvas.us. Never open any other link in the board's browser.

Before every build, call board_status to get the live screen width and height, then build for that size: fill the screen, no scrollbars, scale proportionally, never stretch.

Pick the layout by item count: 1 item = one full-screen picture with a short caption. 2 to 4 items = a two-column grid (grid-2), or a big picture plus up to 3 cards. 5 or 6 items = a three-column grid (grid-3) or a list with small pictures (list-thumbs). More than 6 = push another push_html page (still one page per screenful). Never use four columns.

Every board must include at least one real picture. If you can't find a real picture, use a big title with an icon-like emoji instead of a wall of text; never invent picture links. Use stable public https picture links. For temporary or generated picture links, set save_copy so the board keeps its own copy. Labels are 2 to 4 words, and a screen holds about 30 words at most. push_text is only for a short title or note (up to 140 characters) and never for multi-item answers.

When your assistant asks you to sign in, type a pairing code from your bot, or start a new board. Then say "put this on my board."

Back to BotCanvas

BotCanvas was created by Pawel Raszewski as a personal project, first built for his Tesla Cybertruck in October 2026. Not affiliated with or endorsed by any employer, or by Tesla. © 2026 Pawel Raszewski. All rights reserved for now. About