Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsBlogPro
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Authors
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges
  • Chrome Extension
  • Skill Manager

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

ProTermsPrivacyRefunds
Back to skills

Pick

ASecurity

Pick an element from the browser by clicking it (starts the dev server if needed) and get back its CSS selector, tag, classes, and text. Manual only.

2 stars
0 votes
0 copies
4 views
Added 9/20/2026
ai-agentsgobashdebugging

Works with

cursorcli

Security Analysis

A96/100
mediumUses curl or wget to download content

Pro scans all 2 files and shows the line behind each finding

Scanned 9/20/2026

$npx -y skills add lttr/claude-marketplace --skill pick --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Pick?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Pick
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/lttr-pick/badge)](https://www.skillsdirectory.com/skills/lttr-pick)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
Files
SKILL.md
---
name: pick
description: Pick an element from the browser by clicking it (starts the dev server if needed) and get back its CSS selector, tag, classes, and text. Manual only.
argument-hint: [prompt]
disable-model-invocation: true
allowed-tools: Bash(playwright-cli:*), Bash(google-chrome:*), Bash(open:*), Bash(curl:*), Bash(ss:*), Bash(lsof:*), Bash(seq:*), Bash(sleep:*), Bash(cat:*), Bash(npm:*), Bash(grep:*), Read, Glob
---

# pick

## Context

- Current working directory: !`pwd`
- Package.json dev script (if exists): !`cat package.json 2>/dev/null | grep -E '"dev":|"start":' | head -2 || echo "No dev/start script"`
- Dev host/port hint (from CLAUDE.md, if any): !`grep -rhoE 'https?://[a-zA-Z0-9.-]+:[0-9]+' CLAUDE.md .claude/CLAUDE.md 2>/dev/null | sort -u | head -3 || echo "No CLAUDE.md dev URL hint"`
- LISTEN ports (candidate dev servers): !`{ ss -tlnH 2>/dev/null | awk '{print $4}' || lsof -nP -iTCP -sTCP:LISTEN 2>/dev/null | awk 'NR>1{print $9}'; } | grep -oE ':[0-9]+$' | tr -d ':' | sort -un | grep -vE '^(22|53|631)$' | head -10 || echo "none"`
- Debuggable browser on :9222 (Chrome OR Edge — both CDP): !`curl -s --max-time 1 http://localhost:9222/json/version 2>/dev/null | grep -oE '"Browser": *"[^"]*"' || echo "nothing on :9222"`

## Your task

Help the user pick an element from the browser. `playwright-cli` attaches to
any Chromium browser over CDP (Chrome, Edge, Electron) — there is **no native
picker**, so we inject a JS overlay and poll for the result.

Every command below runs in the named session `-s=pick` (the flag goes before
the command), so it never touches the user's other playwright-cli sessions.

**Step 0: Preflight**

```bash
${CLAUDE_PLUGIN_ROOT}/skills/playwright-cli/check.sh
```

If it fails, stop and relay its message to the user — it says what to install.

**Step 1: Get a debuggable browser**

- **Reuse what's already on :9222** if the context shows one (the user's normal
  Chrome/Edge — best, keeps logged-in sessions). Skip to Step 1b.
- **Else launch a fresh Chrome.** Must be _detached_ (`nohup … &`) or it dies
  when the command returns. `--ignore-certificate-errors --test-type` handles
  flaky dev TLS (e.g. self-signed `*.local` hosts). 9222 is often taken by
  Edge/Teams, so use a free port. The binary name differs per OS: `google-chrome`
  (Linux), `"/Applications/Google Chrome.app/Contents/MacOS/Google Chrome"`
  (macOS), `chrome.exe` or `msedge.exe` (Windows) — Edge takes the same flags.
  ```bash
  nohup google-chrome --remote-debugging-port=9223 \
    --user-data-dir="$HOME/.chrome-debug-profile" \
    --ignore-certificate-errors --test-type "<start-url>" \
    </dev/null >/dev/null 2>&1 &
  until curl -s http://localhost:9223/json/version >/dev/null 2>&1; do sleep 0.3; done
  ```

**Step 1b: Attach + select the tab**

Attach to the browser, list its tabs, and select the one showing the app. Tab
selection is explicit and sticky: every later `eval` hits that tab.

```bash
PORT=9222   # or 9223 if you launched fresh
playwright-cli -s=pick close 2>/dev/null   # drop a stale pick session, if any
playwright-cli -s=pick attach --cdp=http://localhost:$PORT
playwright-cli -s=pick tab-list                 # "- 1: (current) [Title](url)"
playwright-cli -s=pick tab-select <n>
playwright-cli -s=pick --raw eval 'location.href + " | " + document.title'   # confirm right tab
```

**Step 2: Find the page to pick from**

- Prefer the dev URL hint from CLAUDE.md (context above) — it carries the right
  host (some apps key locale/routing off the Host header, so `localhost` ≠ the
  project hostname).
- Else pick the dev server from the LISTEN ports above (ignore system ports).
- If nothing is listening, start the app the way this project starts it: a
  `run` skill for the project if one is available, else whatever CLAUDE.md or
  the README prescribes, else the repo's own package manager and dev script.
  Background it and wait until its port is LISTEN.

**Step 3: Navigate**

```bash
playwright-cli -s=pick goto "<dev-url>"
playwright-cli -s=pick --raw eval 'location.href + " | " + document.title'   # confirm
```

**Step 4: Arm the picker overlay, then poll**

Inject the overlay (multi-select via Ctrl/Cmd+click, Enter finishes, single
click resolves one, ESC cancels):

```bash
playwright-cli -s=pick --raw eval "$(cat "${CLAUDE_SKILL_DIR}/picker.js")"
```

The overlay lives in [`picker.js`](picker.js) next to this file: it highlights
whatever the cursor is over, and on click leaves the pick (selector, text,
truncated HTML, attributes) on `window.__picked` with `window.__pickerDone` set.
Returns `"armed"`, or `"already-armed"` if it is on the page already.

Then poll (eval is one-shot, so loop). `--raw eval` prints the value as JSON,
so returning the object itself gives clean, parseable output:

```bash
for i in $(seq 1 120); do
  r=$(playwright-cli -s=pick --raw eval 'window.__pickerDone ? window.__picked : null' 2>/dev/null)
  [ -n "$r" ] && [ "$r" != "null" ] && { echo "$r"; break; }
  sleep 1
done
[ -z "$r" ] || [ "$r" = "null" ] && echo "TIMEOUT: no element picked in 120s"
```

Use the user's prompt below to tell the user what to click before you start
polling.

**Step 5: Return results**

- If `cancelled`, report the user cancelled.
- For a single pick or each multi entry, show `tag` / `id` / `classes` / `text`
  and the ready-to-use CSS `selector`.
- Detach so the user's browser stays up: `playwright-cli -s=pick detach`.

## User prompt (if any)

$ARGUMENTS

Attribution

lttrlttr
View sourceSee grades on GitHubMore from lttr →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Terse caveman voice: answer first, fluff gone, every technical fact kept. Use for /caveman, "caveman mode", "talk like caveman", "be brief", "less tokens". Stays on until "stop caveman" or "normal mode".

1100021 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

698431 votes

Writing Skills

Create and manage Claude Code skills in HASH repository following Anthropic best practices. Use when creating new skills, modifying skill-rules.json, understanding trigger patterns, working with hooks, debugging skill activation, or implementing progressive disclosure. Covers skill structure, YAML frontmatter, trigger types (keywords, intent patterns), UserPromptSubmit hook, and the 500-line rule. Includes validation and debugging with SKILL_DEBUG. Examples include rust-error-stack, cargo-dep...

3931 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3421 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Amp, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Grok Build, Kimi, OpenCode, Pi Agent, or ZCode session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", asks to recover/summarize a previous session before continuing, or asks to diagnose or report a catchup failure. Do NOT use for the current conversation, git history, or any non-agent log.

741 votes
View all in ai-agents →