Adonis0123/adonis-skills / skills/chrome-dev-mcp
chrome-dev-mcp
VerifiedUse this skill when the user invokes /chrome-dev-mcp or asks for Chrome DevTools MCP, CDP, list_pages/select_page, UXC packaging for Chrome DevTools, DOM snapshots, Console, Network, Performance, Lighthouse, heap analysis, browser-internal debugging, connection recovery, or correct-Chrome validation across Claude Code, Codex, Grok Build/Grok002, Cursor CLI, Hermes, or WorkBuddy. Use it when the evidence is Chrome-specific (Lighthouse, a full trace, heap data) or the caller has no page session yet; a page already open in ego-browser keeps its own Console and request diagnosis. Also use it as the verification step right after the user or this session changes a web UI or asks to accept or check a page: collect a screenshot, Console, and Network evidence before calling it done. Do not use it for ordinary navigation, form filling, scraping, or desktop UI unless browser-internal signals or UI-change acceptance are required. A passing readiness or list_pages check proves the connection only, never the change.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 11.4 KB |
| evals/evals.json | config | 14.7 KB |
| evals/trigger-eval.json | config | 3.3 KB |
| references/host-verification.md | markdown | 5.9 KB |
| references/maintenance.md | markdown | 4.5 KB |
| references/profile-identity.md | markdown | 1.6 KB |
| references/uxc-facade.md | markdown | 7.5 KB |
| scripts/configure-local.zshexecutable | script | 1.8 KB |
| scripts/ensure-connection.zshexecutable | script | 2.6 KB |
| scripts/host-acceptance.zshexecutable | script | 6.9 KB |
| scripts/install-uxc.zshexecutable | script | 4.0 KB |
| scripts/lib/load-config.zshexecutable | script | 438 B |
| scripts/lib/readiness-envelope.zsh | script | 1.5 KB |
| scripts/lib/uxc-link-contract.zsh | script | 1.5 KB |
| scripts/lib/uxc-owned-binary.zsh | script | 2.8 KB |
| scripts/lib/uxc-release.zshexecutable | script | 642 B |
| scripts/setup-uxc-link.zshexecutable | script | 3.1 KB |
| scripts/test/host-acceptance.test.mjs | script | 5.3 KB |
| scripts/test/readiness-envelope.test.mjs | script | 2.7 KB |
| scripts/test/ui-acceptance.test.mjs | script | 5.2 KB |
| scripts/uxc-readiness.zshexecutable | script | 2.7 KB |
scripts/host-acceptance.zsh
View on GitHub#!/bin/zsh
# Run the read-only multi-host acceptance from references/host-verification.md.
# By default each host gets a fresh headless session that may run only this
# skill's readiness helper and shared CLI. --bypass (explicit user approval only)
# gives every host its approve-everything flag instead. --cursor-auto-review
# (explicit user approval only) lets Cursor CLI's server-side classifier approve
# shell calls; without it Cursor's headless default mode refuses them.
set -euo pipefail
skill_dir="${0:A:h:h}"
timeout_s=300
out_dir=""
keep_raw=false
bypass=false
cursor_auto_review=false
typeset -a hosts all_hosts
all_hosts=(claude codex grok cursor)
fail_closed() {
print -u2 -- "STATUS=ERROR"
print -u2 -- "ERROR_CLASS=$1"
exit 69
}
while (( $# > 0 )); do
case "$1" in
--host)
[[ $# -ge 2 && " ${all_hosts[*]} " == *" $2 "* ]] || fail_closed "unknown_host"
hosts+=("$2")
shift 2
;;
--timeout)
[[ $# -ge 2 && "$2" == <-> ]] || fail_closed "bad_timeout"
timeout_s="$2"
shift 2
;;
--out)
[[ $# -ge 2 && -n "$2" ]] || fail_closed "missing_option_value"
out_dir="$2"
shift 2
;;
--keep-raw)
keep_raw=true
shift
;;
--bypass)
bypass=true
shift
;;
--cursor-auto-review)
cursor_auto_review=true
shift
;;
*) fail_closed "unexpected_arguments" ;;
esac
done
(( ${#hosts} > 0 )) || hosts=("${all_hosts[@]}")
[[ -n "$out_dir" ]] || out_dir="$(mktemp -d "${TMPDIR:-/tmp}/chrome-dev-mcp-acceptance.XXXXXX")"
/bin/mkdir -p "$out_dir"
/bin/chmod 700 "$out_dir"
prompt='只读验收:必须通过 chrome-dev-mcp skill 的共享 CLI 真实调用 list_pages 一次;不要导航、点击,也不要输出页面 URL、标题或正文。成功只回答 CHROME_DEV_MCP_READY,共享调用失败只回答 CHROME_DEV_MCP_FAIL。不要自动回退到 native MCP。'
# Commands each host may run, matching the skill's documented entry points.
readiness_rule='zsh scripts/uxc-readiness.zsh'
cli_rule='chrome-dev-mcp-cli'
run_host() {
local host="$1" raw="$out_dir/$1.raw" final="$out_dir/$1.final"
: >| "$final"
case "$host:$bypass" in
claude:false)
claude -p "/chrome-dev-mcp $prompt" \
--output-format stream-json --verbose \
--allowedTools "Skill(chrome-dev-mcp)" "Bash($readiness_rule*)" "Bash($cli_rule *)"
;;
claude:true)
claude -p "/chrome-dev-mcp $prompt" \
--output-format stream-json --verbose --permission-mode bypassPermissions
;;
codex:false)
# workspace-write keeps the sandbox; the shared daemon socket and cache live in ~/.uxc.
codex exec --skip-git-repo-check -C "$skill_dir" \
-s workspace-write -c sandbox_workspace_write.network_access=true \
--add-dir "$HOME/.uxc" -o "$final" \
"\$chrome-dev-mcp $prompt"
;;
codex:true)
codex exec --skip-git-repo-check -C "$skill_dir" \
--dangerously-bypass-approvals-and-sandbox -o "$final" \
"\$chrome-dev-mcp $prompt"
;;
grok:false)
grok -p "/chrome-dev-mcp $prompt" --output-format streaming-json \
--allow "Bash($readiness_rule*)" --allow "Bash($cli_rule*)"
;;
grok:true)
grok -p "/chrome-dev-mcp $prompt" --output-format streaming-json --always-approve
;;
cursor:false)
# Headless default mode rejects shell calls, so the task never reaches the CLI.
# --auto-review cannot be narrowed to the two allowed commands: opt-in only.
if [[ "$cursor_auto_review" == true ]]; then
cursor-agent -p --trust --auto-review --output-format stream-json "$prompt"
else
cursor-agent -p --trust --output-format stream-json "$prompt"
fi
;;
cursor:true)
cursor-agent -p --trust --force --output-format stream-json "$prompt"
;;
esac >| "$raw" 2>&1
}
run_with_timeout() {
local host="$1" pid watchdog rc
( cd "$skill_dir" && run_host "$host" ) &
pid=$!
# Poll instead of one long sleep so no orphaned sleep holds the caller's pipes open.
(
integer waited=0
while kill -0 "$pid" 2>/dev/null && (( waited < timeout_s )); do
sleep 1
(( waited += 1 ))
done
kill -TERM "$pid" 2>/dev/null || true
) >/dev/null 2>&1 &
watchdog=$!
rc=0
wait "$pid" || rc=$?
wait "$watchdog" 2>/dev/null || true
print -r -- "$rc" >| "$out_dir/$host.exit"
}
extract_final() {
local host="$1" raw="$out_dir/$1.raw" final="$out_dir/$1.final"
[[ -s "$final" ]] && return
case "$host" in
claude|cursor)
/usr/bin/jq -rR 'fromjson? | select(.type == "result") | .result // empty' "$raw" >| "$final" 2>/dev/null || true
;;
grok)
# The answer is the text streamed after the last tool call; thoughts may quote the prompt's sentinels.
/usr/bin/jq -rRs '[split("\n")[] | fromjson? | select(type == "object")]
| (map(.type) | rindex("tool_call_update") // -1) as $last
| .[($last + 1):] | map(select(.type == "text") | .data // empty) | join("")' \
"$raw" >| "$final" 2>/dev/null || true
;;
esac
# Fallback: the last sentinel the host printed.
/usr/bin/grep -q "[^[:space:]]" "$final" 2>/dev/null || /usr/bin/grep -oE 'CHROME_DEV_MCP_(READY|FAIL)' "$raw" 2>/dev/null | /usr/bin/tail -1 >| "$final" || true
}
# Private page URLs for the leak check; never printed.
url_file="$out_dir/.page-urls"
chrome-dev-mcp-cli --timeout-ms 20000 list_pages 2>/dev/null |
/usr/bin/jq -r '.data.content[0].text // empty' |
/usr/bin/sed -nE 's/^[0-9]+: .*\(([^()]*)\)( \[selected\])?$/\1/p' >| "$url_file" || true
for host in "${hosts[@]}"; do
if command -v "${${host/cursor/cursor-agent}}" >/dev/null 2>&1; then
run_with_timeout "$host" &
else
print -r -- "127" >| "$out_dir/$host.exit"
: >| "$out_dir/$host.raw"
fi
done
wait
overall=0
for host in "${hosts[@]}"; do
extract_final "$host"
raw="$out_dir/$host.raw"
final="$out_dir/$host.final"
rc="$(<"$out_dir/$host.exit")"
trace=NO
/usr/bin/grep -qE 'uxc-readiness\.zsh|chrome-dev-mcp-cli' "$raw" 2>/dev/null && trace=YES
sentinel=NONE
if /usr/bin/grep -q 'CHROME_DEV_MCP_FAIL' "$final" 2>/dev/null; then
sentinel=FAIL
elif /usr/bin/grep -q 'CHROME_DEV_MCP_READY' "$final" 2>/dev/null; then
sentinel=READY
fi
leak=NO
if [[ -s "$url_file" ]] && /usr/bin/grep -qFf "$url_file" "$final" 2>/dev/null; then
leak=YES
fi
result=UNVERIFIED
if [[ "$rc" == 127 ]]; then
result=HOST_MISSING
elif [[ "$sentinel" == READY && "$trace" == YES && "$leak" == NO && "$rc" == 0 ]]; then
result=VERIFIED
elif [[ "$sentinel" == FAIL || "$leak" == YES ]]; then
result=FAIL
fi
[[ "$result" == VERIFIED ]] || overall=1
print -r -- "HOST=$host RESULT=$result SENTINEL=$sentinel TOOL_TRACE=$trace LEAK=$leak EXIT=$rc"
done
/bin/rm -f -- "$url_file"
if [[ "$keep_raw" == true ]]; then
print -r -- "RAW_DIR=$out_dir"
else
/bin/rm -f -- "$out_dir"/*.raw
fi
print -r -- "OUT_DIR=$out_dir"
exit "$overall"
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| brainstormingobra/superpowersYou MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation). | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| brand-guidelinesanthropics/skillsApplies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| frontend-designanthropics/skillsGuidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| mcp-builderanthropics/skillsGuide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK). | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |