How do I install the Cmux Browser Screenshooting skill?

Run npx skillmds@latest add erich3000/cmux-browser-screenshooting in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Cmux Browser Screenshooting safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Cmux Browser Screenshooting?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Cmux Browser Screenshooting free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Cmux Browser Screenshooting?

erich3000 (@erich3000) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Cmux Browser Screenshooting by erich3000

Name: Cmux Browser Screenshooting
Author: erich3000

Cmux Browser Screenshooting

This skill should be used when the user asks to "take a screenshot", "screenshot the browser", "show me the browser", "capture the screen", "screenshot storybook", "open storybook in browser", "navigate to URL", "open URL in cmux", or wants to visually verify what is currently shown in a browser, open a new browser pane, or navigate an existing browser to a specific URL before screenshotting. Always use this skill instead of mcp__chrome-devtools__take_screenshot — cmux is token-efficient.

erich3000 2008199 2.8 KB Updated Sep 17, 2026

File contents

cmux Browser Screenshooting

Take a screenshot of a browser pane in the current cmux workspace. Can open a new browser, navigate an existing one to a URL, or screenshot whatever is currently shown.

Step 1 — Discover workspace and browser surface

Run both commands, then parse:

WS=$(cmux identify | jq -r '.focused.workspace_ref')
SURF=$(cmux rpc surface.list "{\"workspace_ref\":\"$WS\"}" | jq -r '[.surfaces[] | select(.type == "browser")] | first | .ref // empty')

If SURF is non-empty → a browser pane exists. Check its URL if needed:
```
cmux browser $SURF url
```
If SURF is empty → no browser pane open.

Step 2 — Open or navigate

No browser exists → open one

cmux browser open <url>

Then re-discover the surface ref (repeat Step 1) since the new surface needs a ref.

Use open-split instead of open to open alongside the current terminal without replacing it:

cmux browser open-split <url>

Browser exists but wrong URL → navigate it

cmux browser $SURF navigate <url>

Wait for the page to load before screenshotting:

cmux browser $SURF wait --load-state complete

Browser exists with correct URL → proceed to screenshot

No action needed.

Step 3 — Take the screenshot

cmux browser $SURF screenshot --out /tmp/cmux-shot.png

Step 4 — Display it

Read /tmp/cmux-shot.png with the Read tool to render the image inline.

Full one-liner (screenshot existing browser)

WS=$(cmux identify | jq -r '.focused.workspace_ref') && \
SURF=$(cmux rpc surface.list "{\"workspace_ref\":\"$WS\"}" | jq -r '[.surfaces[] | select(.type == "browser")] | first | .ref') && \
cmux browser $SURF screenshot --out /tmp/cmux-shot.png

Notes

Never use mcp__chrome-devtools__take_screenshot — it is token-expensive.
If multiple browser surfaces exist, pick the one whose title matches the relevant page (e.g. "Storybook"). Use jq -r '.surfaces[] | select(.type == "browser") | [.ref, .title] | @tsv' to list all browser surfaces with titles.
Surface refs change between sessions — always discover fresh, never hardcode.
Prefer open-split over open to keep the terminal pane visible alongside the browser.

---name: cmux-browser-screenshootingdescription: This skill should be used when the user asks to "take a screenshot", "screenshot the browser", "show me the browser", "capture the screen", "screenshot storybook", "open storybook in browser", "navigate to URL", "open URL in cmux", or wants to visually verify what is currently shown in a browser, open a new browser pane, or navigate an existing browser to a specific URL before screenshotting. Always use this skill instead of mcp__chrome-devtools__take_screenshot — cmux is token-efficient.---# cmux Browser ScreenshootingTake a screenshot of a browser pane in the current cmux workspace. Can open a new browser, navigate an existing one to a URL, or screenshot whatever is currently shown.## Step 1 — Discover workspace and browser surfaceRun both commands, then parse:```bashWS=$(cmux identify | jq -r '.focused.workspace_ref')SURF=$(cmux rpc surface.list "{\"workspace_ref\":\"$WS\"}" | jq -r '[.surfaces[] | select(.type == "browser")] | first | .ref // empty')```- If `SURF` is non-empty → a browser pane exists. Check its URL if needed: ```bash cmux browser $SURF url ```- If `SURF` is empty → no browser pane open.## Step 2 — Open or navigate### No browser exists → open one```bashcmux browser open <url>```Then re-discover the surface ref (repeat Step 1) since the new surface needs a ref.Use `open-split` instead of `open` to open alongside the current terminal without replacing it:```bashcmux browser open-split <url>```### Browser exists but wrong URL → navigate it```bashcmux browser $SURF navigate <url>```Wait for the page to load before screenshotting:```bashcmux browser $SURF wait --load-state complete```### Browser exists with correct URL → proceed to screenshotNo action needed.## Step 3 — Take the screenshot```bashcmux browser $SURF screenshot --out /tmp/cmux-shot.png```## Step 4 — Display itRead `/tmp/cmux-shot.png` with the Read tool to render the image inline.## Full one-liner (screenshot existing browser)```bashWS=$(cmux identify | jq -r '.focused.workspace_ref') && \SURF=$(cmux rpc surface.list "{\"workspace_ref\":\"$WS\"}" | jq -r '[.surfaces[] | select(.type == "browser")] | first | .ref') && \cmux browser $SURF screenshot --out /tmp/cmux-shot.png```## Notes- Never use `mcp__chrome-devtools__take_screenshot` — it is token-expensive.- If multiple browser surfaces exist, pick the one whose `title` matches the relevant page (e.g. "Storybook"). Use `jq -r '.surfaces[] | select(.type == "browser") | [.ref, .title] | @tsv'` to list all browser surfaces with titles.- Surface refs change between sessions — always discover fresh, never hardcode.- Prefer `open-split` over `open` to keep the terminal pane visible alongside the browser.

Frequently asked questions