Direct Chrome control through the use_browser MCP tool: tabs, forms, content extraction
One action-based MCP tool covering navigate/click/type/extract/eval, multi-tab management, dialog handling, console capture, auto-capture of screenshots and DOM, and credential-shaped content redaction
- What
- One action-based MCP tool covering navigate/click/type/extract/eval, multi-tab management, dialog handling, console capture, auto-capture of screenshots and DOM, and credential-shaped content redaction
- Cost
- Free
- Needs
- obra's chrome-ws bridge exposing the mcp__chrome__use_browser MCP tool in your runtime (for example Claude Code) — this skill teaches the protocol; the driver itself must already be installed.
- Install
- Copy the installer prompt below into your Muse — your agent does the rest.
Curated by Skill Harbor — @obra's browsing skill: full control of a real Chrome through a single `use_browser` MCP tool — navigate and wait (elements, text), click/type/select/keyboard interactions, CDP-level mouse (hover, drag-drop, move, scroll) that bypasses synthetic-event detection, file uploads, extraction (markdown/text/html/attributes), `eval` for arbitrary JavaScript, screenshots, multi-tab management with sticky active tabs, headed/headless mode control, profile management with multi-MCP disambiguation, kill/restart lifecycle with auto-restart detection, console logging per tab, and native dialog handling (alerts, basic-auth, permissions, device choosers). Thoughtful safety design: every DOM action auto-captures a screenshot, structured markdown and the DOM for inspection; credential-shaped pages suppress capture and redact secrets in output (overridable via `SUPERPOWERS_CHROME_ALLOW_CREDENTIAL_CAPTURE=1` — leave it off); mode toggles restart Chrome and lose POST state, so they are warned loudly. Honest caveats: it requires the `mcp__chrome__use_browser` MCP server (obra's chrome-ws bridge) installed in the agent's runtime — the skill teaches the protocol, it doesn't install the driver; Playwright MCP remains the better pick for fresh throwaway instances and screenshots/PDFs. MIT licensed. Skill Harbor never reviews the code, review it yourself before use. Discovered via skills.sh.
Version:
Quick answer
Direct Chrome control is a free Muse skill giving full control of real Chrome through one action-based MCP tool, use_browser. It covers navigate, click, type, extract and eval, multi-tab management, dialog handling, console capture and auto-capture of screenshots and DOM. Credential-shaped content is redacted, keeping sensitive-looking data out of captured output.
Install
Copy the install package below, then paste it into Muse🧪 Install ran in our test sandbox · Nov 11, 2026📌 Pinned version: commit eabb66fBefore you start: tell me in a few lines what this skill does and what installing it involves, and wait for my go-ahead. Prerequisites: obra's chrome-ws bridge exposing the mcp__chrome__use_browser MCP tool in your runtime (for example Claude Code) — this skill teaches the protocol; the driver itself must already be installed. Install "Direct Chrome control through the use_browser MCP tool: tabs, forms, content extraction" for me. It gives you obra's browsing playbook: one action-based MCP tool for navigate/await/click/type/keyboard/drag-drop/scroll/file-upload/extract/eval in Chrome, multi-tab management, dialog handling (alerts, basic-auth, permissions), console capture, a screenshot and DOM snapshot saved after every action, profile and lifecycle management, and common patterns (form filling, multi-tab workflows, dynamic content, responsive testing). It guards secrets as it goes: credential-shaped content is redacted from what it saves and returns, and captures are suppressed on pages it treats as sensitive. MIT licensed. Source: https://raw.githubusercontent.com/obra/superpowers-chrome/eabb66fabd244cd6eaddc13a652191179fda3a99/skills/browsing/SKILL.md (the helper files live in the same folder, skills/browsing) Pinned version: install the code at commit eabb66fabd244cd6eaddc13a652191179fda3a99 exactly (Skill Harbor tested this version). Do not substitute a newer version. If this commit is unreachable, stop and tell me. Updated October 11, 2026: Skill Harbor regenerated this install prompt from the upstream file of the day. What changed for you: the skill's rules for keeping secrets out of extract and eval results were rewritten. The list of pages the skill treats as sensitive grew — it now names Slack's App-Level Tokens page and GitHub's, Google's and Linear's own token pages — and extracting a whole page on those pages still refuses outright. The checks now work like this: a complete token shape and a dense list of secret-shaped codes are refused on any page, while a bare token prefix and a mixed letter-and-digit run are only checked on the listed sensitive pages. On a listed page, eval now runs and its result is checked against those same rules before it comes back, instead of refusing outright as before — and credential-shaped text is still masked out of anything the skill returns, on any page. The previous prompt described the older rules. 1. Download SKILL.md and the helper files from the same folder (skills/browsing) into a temporary folder. 2. Compare SKILL.md with the fingerprint from Skill Harbor's review of 2026-10-11: - SKILL.md: SHA-256 79e503b283e6414046bf639d227f1b1d7180dbb04a48b03a095abc402c9b0aa5 If it does not match, the file changed since the review: do not install it yet — tell me what you found and let me decide. 3. Read the file and tell me in plain words: what the skill does when used, whether it connects to a service, whether it reads or sends any personal information or keys, and anything that seems unusual. 4. When I say yes, copy SKILL.md and its helper files into my skills folder, in a folder named "browsing". 5. Check the copy opens correctly (valid frontmatter, files in place) and tell me what was installed, where, and what is left for me to do (for example: confirm the use_browser MCP tool is available in your runtime and announce its use as the skill asks; keep SUPERPOWERS_CHROME_ALLOW_CREDENTIAL_CAPTURE unset — setting it turns the secret protections off; remember show_browser/hide_browser restarts Chrome and loses POST state, so prefer headless). GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Rules: only touch the temporary folder and the install target. If something seems off, stop and ask me.
Saved to your recent installs. Find it anytime on /connect.
🔒 Security, provenance and what this touches
The install prompt below already includes the vetting steps: your agent follows the community checklist before installing anything with executable code. Want more?
Verification certificate
Dated facts about what was checked for this listing, and how. This is not a safety certification.
- Pinned version
- eabb66f · 2026-10-13
- Sandbox install test
- Install ran in our test sandbox · Nov 11, 2026 · Observed during the install test: registry.npmjs.org
- Creator identity
- Not verified
- Fingerprint watch
- Last checked · Nov 19, 2026 : no change detected in the files linked by this install prompt
- Install prompt analysis
- 2026-10-13 : No commands, network calls, packages, file writes or API keys detected.
Skill Harbor never reviews the code of a listed skill. These are dated facts about the install prompt and our install test, not a safety certification. Review a skill yourself before you install it. How we verify
Questions
How do I install a skill?
Every product page includes a copy-paste install prompt. Paste it into your Muse and it sets the skill up for you — no manual configuration.
Where does my money go?
Straight to the seller. Skill Harbor never processes payments: checkout happens on the seller’s own page, usually Stripe.
What does the ✓ next to a creator’s name mean?
It means we confirmed the identity of the person behind the listing. It says nothing about the code itself — always check a skill before installing it.