Browser Use
LeoYeAI/openclaw-master-skills
Automates browser interactions for social media management across Instagram, LinkedIn, and X.
Recon-first browser automation. An agent skill from team-attention/hoyeon.
Install Browser Work
$ npx skills add team-attention/hoyeon --skill browser-work -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install team-attention/hoyeon browser-work --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/team-attention/hoyeon.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/browser-work .claude/skills/browser-work && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "browser-work" agent skill from https://github.com/team-attention/hoyeon/tree/main/skills/browser-work into .claude/skills/browser-work/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "browser-work", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/team-attention/hoyeon/tree/main/skills/browser-workType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add team-attention/hoyeon --skill browser-work -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install team-attention/hoyeon browser-work --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/team-attention/hoyeon.git skills-src && mkdir -p .agents/skills && cp -r skills-src/skills/browser-work .agents/skills/browser-work && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "browser-work" agent skill from https://github.com/team-attention/hoyeon/tree/main/skills/browser-work into .agents/skills/browser-work/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "browser-work", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add team-attention/hoyeon --skill browser-work -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install team-attention/hoyeon browser-work --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/team-attention/hoyeon.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/skills/browser-work .cursor/skills/browser-work && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "browser-work" agent skill from https://github.com/team-attention/hoyeon/tree/main/skills/browser-work into .cursor/skills/browser-work/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "browser-work", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/team-attention/hoyeon.git --path skills/browser-work--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add team-attention/hoyeon --skill browser-work -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install team-attention/hoyeon browser-work --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/team-attention/hoyeon.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/skills/browser-work .gemini/skills/browser-work && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "browser-work" agent skill from https://github.com/team-attention/hoyeon/tree/main/skills/browser-work into .gemini/skills/browser-work/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "browser-work", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install team-attention/hoyeon browser-workInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add team-attention/hoyeon --skill browser-work -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/team-attention/hoyeon.git skills-src && mkdir -p .github/skills && cp -r skills-src/skills/browser-work .github/skills/browser-work && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "browser-work" agent skill from https://github.com/team-attention/hoyeon/tree/main/skills/browser-work into .github/skills/browser-work/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "browser-work", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add team-attention/hoyeon --skill browser-work -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install team-attention/hoyeon browser-work --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/team-attention/hoyeon.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/skills/browser-work .opencode/skills/browser-work && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "browser-work" agent skill from https://github.com/team-attention/hoyeon/tree/main/skills/browser-work into .opencode/skills/browser-work/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "browser-work", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Facts
- Skill name
browser-work- GitHub stars
- 173
- Token cost
- ~1.9k tokens
- SKILL.md length
- 684 words
- Files
- 2 (incl. references)
- Skills in repo
- 36
- Repo updated
- First seen
- Licence
- MIT
At a glance
Recon-first browser automation. An agent skill from team-attention/hoyeon.
- Works in 6 steps: Setup → Assess Complexity → Recon (Orchestrator explores directly) → …
- : /browser-work
- SKILL.md covers Purpose, Runtime Surface, Why Recon First? and Execution, plus 2 more sections
- Calls npx, openssl and npm
What it does
Browser Work is an agent skill from team-attention/hoyeon. Recon-first browser automation. Orchestrator explores the site first via chromux, saves a guide file with insights, then delegates execution to browser-explorer agent. Use when: "/browser-work", "브라우저 작업", "사이트에서 해줘", "웹에서 해줘", "LinkedIn에서", "크롬으로", "browser task", "automate this site".
Its SKILL.md is about 1.9k tokens, which your agent loads only when the skill is triggered. The skill folder holds 2 other files, including reference files (for example `references/chromux-guide.md`).
It sits in Productivity & Automation, covering Browser automation. It works with LinkedIn. The repository describes itself as: Requirements-first Harness — derive, verify, execute. The licence is MIT.
When your agent uses it
- : /browser-work
- Automate this site
Example prompts
- “/browser-work”
- “LinkedIn에서”
- “browser task”
- “/browser-work”
Requirements
- Node.js
Workflow steps
6 steps, taken from the step headings in SKILL.md.
What it can do on your machine
Read from SKILL.md and the folder at commit 7cff032. It shows what the files ask for, not the result of running them.
Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
Shell commands in SKILL.md call:
npxopensslnpm
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md. Its commands use npx and npm, which can reach the network depending on how they are called.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Context cost
Browser Work loads about 1.9k tokens when it runs, and up to ~2.7k if it reads all its reference files. Until then it costs about 75 tokens; SKILL.md has 684 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
Safety
Auto-check passed
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.
SKILL.md
The full file from team-attention/hoyeon at commit 7cff032, republished under its MIT licence (© team-attention). 684 words, ~1,869 tokens.
Download SKILL.mdSave it as
.claude/skills/browser-work/SKILL.md (or your agent's skills folder). This skill also uses 1 other file; get the full folder from GitHub.- name
- browser-work
- description
- Recon-first browser automation. Orchestrator explores the site first via chromux, saves a guide file with insights, then delegates execution to browser-explorer agent. Use when: "/browser-work", "브라우저 작업", "사이트에서 해줘", "웹에서 해줘", "LinkedIn에서", "크롬으로", "browser task", "automate this site".
- version
- 1.0.0
Browser Work
Recon-first browser automation: explore → document → delegate.
Purpose
To reliably execute browser tasks on the user's behalf, the orchestrator first scouts the site directly, creates a pitfall-prevention guide, then delegates execution to the browser-explorer agent.
Runtime Surface
Claude Code
- Use hook-provided
CLAUDE_SESSION_IDfor~/.hoyeon/{session}/paths. - Delegate with the logical browser agent described below.
Codex
- Use Bash-first chromux operations. Do not add Hoyeon MCP for v1.
- If no hook-provided session ID exists, generate one with
date +%Y%m%d-%H%M%Sand store guides under$HOME/.hoyeon/codex-browser-$RUN_ID/. - Map the logical browser agent to the Codex adapter
hoyeon-browser-explorerwhen installed. - If the current Codex session has not loaded that adapter, complete the smallest safe browser pass directly with chromux and report the fallback.
- Use the Browser Use plugin only when the user explicitly asks for Codex in-app browser behavior; this skill's canonical browser runtime is chromux.
Why Recon First?
When an agent sees a site for the first time, there's a lot of trial and error (snapshot vs. screenshot confusion, clicking wrong elements, unfamiliar site structure). If the orchestrator walks through one cycle first and builds a "map," the agent can execute accurately.
Execution
Step 0: Setup
0-1. Session Init
SESSION_ID="[CLAUDE_SESSION_ID from UserPromptSubmit hook]"
WORK_DIR="$HOME/.hoyeon/$SESSION_ID"
mkdir -p "$WORK_DIR"
echo "WORK_DIR=$WORK_DIR"0-2. Chromux Check
Resolve chromux path. Remember the output literally — you'll inline it in every command.
CX=$(command -v chromux 2>/dev/null || echo "") && [ -n "$CX" ] && echo "CHROMUX=$CX" || (npx @team-attention/chromux help >/dev/null 2>&1 && echo "CHROMUX=npx @team-attention/chromux" || echo "MISSING")If MISSING, report error and stop.
Launch Chrome in headless mode (no visible window, but fully functional):
/path/to/chromux launch default --headless 2>/dev/null || trueTo let the user see a live tab (e.g., during recon or debugging), use show — no restart needed:
/path/to/chromux show exp-ab12 # Opens DevTools in user's browser0-3. Generate Session ID
openssl rand -hex 2Remember the output (e.g., ab12) → your chromux session ID is exp-ab12. Inline it literally in every command.
Step 1: Assess Complexity
Before doing recon, assess whether the task needs it:
| Complexity | Criteria | Action |
|---|---|---|
| Simple | Single page, 1-2 clicks, well-known site (Google, GitHub) | Skip recon → go directly to Step 4 (Delegate) |
| Medium | Multi-step workflow, unfamiliar site, 3+ interactions | Do recon (Step 2-3) |
| Complex | Dynamic content, auth flows, pagination, bot-sensitive site | Do thorough recon (Step 2-3) + extra caution notes |
If skipping recon, still create a minimal guide file with the task description and URL.
Step 2: Recon (Orchestrator explores directly)
You (the orchestrator) use chromux directly. Follow the chromux guide in references/chromux-guide.md.
2-1. Navigate & Snapshot
/path/to/chromux open exp-ab12 "<target-url>" && sleep 2 && /path/to/chromux snapshot exp-ab12Show full SKILL.md (322 more words)Show less
2-2. Walk Through the Workflow
Execute the entire workflow once — the same steps the agent will need to do:
- Snapshot the page → identify key elements and their @ref numbers
- Click/interact as needed → observe what changes
- Snapshot again after each action → note how @ref numbers shift
- Note obstacles: popups, modals, login walls, infinite scroll, dynamic loading
- Note patterns: does "load more" change @ref numbers? Are there confirmation dialogs?
Bot detection caution:
- Add
wait 2000between actions (don't click rapidly) - Don't repeat the same action more than 3 times quickly
- If you see a CAPTCHA or rate limit warning, stop and note it in the guide
2-3. Close Recon Session
/path/to/chromux close exp-ab12Step 3: Write Guide File
Save recon findings to $WORK_DIR/guide.md. This is the "map" the agent will follow.
cat > "$WORK_DIR/guide.md" << 'GUIDE_EOF'
# Browser Work Guide
## Task
[What the user wants done — 1-2 sentences]
## Target URL
[Starting URL]
## Site Characteristics
- [Login required? Already logged in?]
- [Single page or multi-page workflow?]
- [Dynamic content loading? (infinite scroll, AJAX)]
- [Known bot detection? Rate limits?]
## Workflow Steps
1. [Step description] — [which element to look for in snapshot]
2. [Step description] — [expected @ref pattern or text to search for]
3. ...
## Pitfalls & Insights
- [Things that could trip up the agent]
- [e.g., "Sort dropdown is NOT the first artdeco-dropdown — look for text 'Most Relevant'"]
- [e.g., "'Load more' button changes @ref every time — always re-snapshot"]
- [e.g., "Confirmation modal appears after clicking Connect — look for 'Send without a note'"]
## Bot Detection Notes
- [Recommended delay between actions]
- [Any rate limits observed]
- [Pages to avoid rapid-fire clicking on]
GUIDE_EOFFill in the template with actual findings from your recon. Be specific — the agent will read this literally.
Step 4: Delegate to Browser-Explorer Agent
Launch the browser-explorer agent with the guide file content included in the prompt.
Agent(
subagent_type: "hoyeon:browser-explorer",
mode: "dontAsk",
prompt: """
[Task description from user]
## Recon Guide
[Paste full contents of $WORK_DIR/guide.md here]
## Execution Rules
1. Follow the Workflow Steps in the guide above
2. Be conservative — add `wait 2000` between actions to avoid bot detection
3. If something doesn't match the guide (unexpected popup, different layout), snapshot and adapt
4. If you hit a CAPTCHA or rate limit, STOP and report back
5. Close your session when done
"""
)If the task requires multiple independent sub-tasks (e.g., "send connection requests to 5 people"), you can launch multiple browser-explorer agents in parallel — each gets its own tab.
Step 5: Report Results
After the agent completes:
- Summarize what was accomplished
- Note any failures or partial completions
- If the agent hit issues not covered by the guide, update
$WORK_DIR/guide.mdwith new insights for future runs
Error Handling
| Situation | Response |
|---|---|
| chromux not found | Report error, suggest npm i -g @team-attention/chromux |
| Site requires login | Check if chromux profile has saved login. If not, tell user to log in manually first |
| CAPTCHA during recon | Stop recon, note in guide, delegate with extra caution |
| Agent fails despite guide | Resume agent with corrections, or re-do recon with more detail |
| Task too complex for single agent | Split into sub-tasks, delegate each to separate agent |
Cleanup
Guide files persist in ~/.hoyeon/{sid}/ for reference. No auto-cleanup — user can review or reuse.
© team-attention, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Files
SKILL.md and 1 other file (references) in skills/browser-work of team-attention/hoyeon.
- SKILL.md
- references/chromux-guide.md
Open the folder on GitHubat commit 7cff032
Compare with similar skills
Browser Work next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Browser Work this skillteam-attention/hoyeon | 173 | — | ~1.9k | Automated safety check: Pass | MIT | |
| Browser UseLeoYeAI/openclaw-master-skills | 2.2k | — | ~4.4k | Automated safety check: Pass | MIT | |
| Browser Automation Edge Casesaden-hive/hive | 11k | — | ~1.8k | Automated safety check: Pass | MIT | |
| Actionbookactionbook/actionbook | 1.6k | — | ~1.5k | Automated safety check: Pass | Apache-2.0 | |
| Bright Data MCPbrightdata/skills | 264 | 1 repos | ~3.7k | Automated safety check: Pass | MIT | |
| Website Browsing Skills Indexbrowsing-skills/browsing-skills | 116 | — | ~1.6k | Automated safety check: Pass | MIT |
Similar skills
Browser Automation Edge Cases
aden-hive/hive
Step-by-step procedure for debugging browser automation failures on complex sites such as LinkedIn, Twitter/X, single-page apps and Shadow DOM pages.
Actionbook
actionbook/actionbook
Activate when the user needs to interact with any website — browser automation, web scraping, screenshots, form filling, UI testing, monitoring, or building AI agents.
Bright Data MCP
brightdata/skills
Bright Data MCP handles ALL web data operations. An agent skill from brightdata/skills.
Website Browsing Skills Index
browsing-skills/browsing-skills
Umbrella skill for a library of website-specific browsing skills. Use when the user's request targets one of these specific websites: <!-- DOMAINS:START…
Non-Destructive Chrome Automation
glebis/claude-skills
Automates a dedicated, logged-in Chrome instance per profile without ever closing the user's own open tabs or browser windows.
More from team-attention/hoyeon
All 36 skills in this repoSkill Session Analyzer
team-attention/hoyeon
This skill should be used when the user asks to "analyze session", "evaluate skill execution", "check session logs", provides a session ID with a skill path, or wants to verify that a skill executed…
Check
team-attention/hoyeon
This skill should be used when the user wants to verify their changes before pushing, or update the project's rule checklists.
Compound
team-attention/hoyeon
This skill should be used when the user says "/compound", "compound this", "document learnings", "save what we learned", or after completing a PR.
QA
team-attention/hoyeon
Systematically QA test any application — web apps, native macOS apps, Electron apps, CLI tools, interactive REPLs, or anything on screen.
Dev Scan
team-attention/hoyeon
Collect diverse opinions on technical topics from developer communities.
Tech Decision
team-attention/hoyeon
This skill should be used when the user asks about "technical decision", "what to use", "A vs B", "comparison analysis", "library selection", "architecture decision", "which one to use"…
Related
Works with
Categories
Questions about Browser Work
What does Browser Work do?
Recon-first browser automation. An agent skill from team-attention/hoyeon. Browser Work is an agent skill from team-attention/hoyeon. Recon-first browser automation.
When should I use Browser Work?
Browser Work fits situations like: : /browser-work; automate this site.
How do I install Browser Work in Claude Code?
Run `npx skills add team-attention/hoyeon --skill browser-work -a claude-code`. Or copy the skill folder (skills/browser-work in team-attention/hoyeon) into .claude/skills/browser-work in your project. Claude Code loads it when a task matches its description.
How do I install Browser Work in Codex?
Run `npx skills add team-attention/hoyeon --skill browser-work -a codex`. Or copy the skill folder (skills/browser-work in team-attention/hoyeon) into .agents/skills/browser-work in your project. Codex loads it when a task matches its description.
Can I use Browser Work in Cursor, Gemini CLI or GitHub Copilot?
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add team-attention/hoyeon --skill browser-work -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/browser-work, .gemini/skills/browser-work, .github/skills/browser-work and .opencode/skills/browser-work in your project.
What does Browser Work need to run?
Going by SKILL.md and its folder, Browser Work needs the command-line tools its instructions call (npx, openssl and npm). Our summary lists: Node.js.
Does Browser Work access the network?
SKILL.md contains no URLs. Its commands use npx and npm, which can reach the network depending on how they are called. This is read from the text; nothing was executed.
Is Browser Work safe to install?
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.
What licence does Browser Work use?
Browser Work is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does Browser Work use?
About 1.9k tokens (SKILL.md is roughly 7.5k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 790 tokens, read only when the agent opens those files.
What are the alternatives to Browser Work?
Skills that share tags, products or a category with Browser Work: Browser Use (LeoYeAI/openclaw-master-skills, 2.2k stars), Browser Automation Edge Cases (aden-hive/hive, 11k stars), Actionbook (actionbook/actionbook, 1.6k stars) and Bright Data MCP (brightdata/skills, 264 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains Browser Work?
team-attention (a GitHub organization) maintains it in team-attention/hoyeon, which has 173 GitHub stars. The repository holds 36 skills in this directory. The repository was last updated on May 21, 2026.
Source: team-attention/hoyeon on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.