dev-browser

An agent skill by SawyerHood, from SawyerHood/dev-browser. Tags: automation, cli, developer-tools, testing, web.

What it does

Browser automation with persistent named pages via the dev-browser CLI. Use when users ask to navigate websites, fill forms, take screenshots, extract web data, test web apps, log into sites, or automate browser workflows. Trigger phrases include "go to [url]", "click on", "fill out the form", "take a screenshot", "scrape", "automate", "test the website", "log into", "open the browser", or any browser interaction request.

Install

With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:

npx skills add SawyerHood/dev-browser --skill dev-browser

Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):

git clone --depth 1 https://github.com/SawyerHood/dev-browser
cp -r dev-browser/skills/dev-browser ~/.claude/skills/dev-browser

Safety box score

Grade D, elevated risk: 49 of 100 risk points, rated 2026-09-23. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.

This skill automates a real Chrome browser with Puppeteer-style JavaScript snippets and persistent named pages. The main risks are arbitrary website interaction, reuse of logged-in browser sessions, and installing an external CLI/binary globally from npm.

Category levels, from 0 (none) to 3 (high)
Category	Level	Why
Code & shell execution	2 (moderate)	It explicitly runs JavaScript snippets through the CLI, e.g. `dev-browser <<'EOF' ...` and `dev-browser -e '...'`, so the agent can execute code locally in the browser-automation context.
Network access	2 (moderate)	The skill is meant to `goto` arbitrary URLs, `fill forms`, `scrape`, and `log into sites`, which means it can send data to websites the user points it at.
Filesystem reach	1 (low)	It can read/write files relative to the current directory for browser uploads and outputs: `file paths resolve against your cwd (uploadFile, screenshot/pdf path)`.
Secrets & credentials	2 (moderate)	Because it controls `a real Chrome` and `log into sites` with `named pages persist between runs`, it can interact with existing browser sessions/cookies in that Chrome profile.
Privilege & persistence	2 (moderate)	Installation is system/user wide via `npm install -g dev-browser` and `dev-browser install`, which changes the user's environment and may fetch a browser binary.
Instruction hijack surface	0 (none)	The skill text does not tell the agent to ignore instructions, hide actions, or follow remote content; it is straightforward browser-automation guidance.
Irreversible actions	1 (low)	It can automate browser workflows like `navigate websites`, `fill forms`, and `log into sites`, but it does not itself request unattended posting, payments, or destructive actions.
Transparency	2 (moderate)	The repo ships only markdown, while the actual functionality comes from the external `dev-browser` package and a possible first-run downloaded binary: `the first run downloads the binary`.

Before installing

Check which Chrome/profile `dev-browser` will use so you know what logged-in sessions it can see.
Install it in a scratch account first and confirm whether `npm install -g` or `dev-browser install` fetches any unreviewed binary.
Run `dev-browser --help` and inspect the installed package version/source before letting it access sensitive sites.

Rated 2026-09-23 by an automated review of 1 file from the skill's repository.

Source

Repository: SawyerHood/dev-browser (all skills from this repository)
Path: skills/dev-browser/SKILL.md
Branch: main
Updated: 2026-09-19

cli-developer — Use when building CLI tools, implementing argument parsing, or adding interactive prompts.
playwright-expert — Use when writing E2E tests with Playwright, setting up test infrastructure, or debugging flaky browser tests.
TUI Puppeteering with tmux — Use when automating or testing TUI/CLI applications - provides isolated tmux sessions with scripts for input, output capture, and state verification
aflpp — Sets up and runs AFL++ for multi-core fuzzing of C/C++ projects built with afl-clang-fast or afl-gcc-fast.
cli-for-agents — Designs or reviews CLIs so coding agents can run them reliably: non-interactive flags, layered --help with examples, stdin/pipelines, fast actionable errors.
control-cli — Build or adapt a local harness to drive, inspect, and profile an interactive CLI or TUI without external services.