claude-code-eyes

# claude-code-eyes — let Claude look at the real world Some outputs live where no test, log, or API can reach them: a display panel, an LED, a breadboard, a rack of wires. This skill grabs the current camera frame and lets you `Read` it, so you answer from what you actually see — not from what the code *should* produce. ## How 1. Run: `bash <path-to-skill>/snap.sh` — it prints the path(s) of the captured JPEG/PNG. (It works from any directory.) 2. `Read` each printed path — the Read tool renders images visually. 3. Answer from the frame. Reference positions concretely: "the red wire on the left rail, third hole down", not "a wire near the top". For **"watch this"** / an action in progress: `bash <path-to-skill>/snap.sh 3 2` (3 frames, 2 s apart), then compare the frames and narrate what changed. Longer processes: raise the count/interval. ## Visual-verify: catch what green tests cannot A rendered screen is an output no unit test can see. This is the use case that makes this more than a webcam grab. **Workflow for any display/render change:** 1. **Before/after** — snap once before the change, once after, and diff the two frames. 2. **Against the source of truth** — take what the spec / API / expected string says the screen *should* show, and check the frame against it character by character, element by element. 3. **Report mismatches** — say exactly what differs, where. **A small case library** (real bugs that every automated layer passed): - **Font cove...