Skills/npm/ai-tag-debugger

ai-tag-debugger@0.22.8

VERIFIED· startup only

Publisher claims: "Open-source AI agent that QA-tests your website's analytics, ad and consent tracking (GA4, GTM, Google Ads, Meta, TikTok). Bring your own model (Claude, OpenAI, Gemini) or use it from your chat app over MCP." Observed in sandbox: no network, touches nothing outside the project, writes no memory.

Registry npm (MCP server)Publisher analitikisler · publishing for 41 daysInstalls 6,146Last detonated 5 h ago
$claude mcp add ai-tag-debugger -- npx -y @analitikisler/ai-tag-debugger
WatchReport an issue

Coverage

Startup only; authenticated tool behavior not tested.
Tested
server startup over stdio (initialize sent) · every file read, file written and host contacted by what ran
Not tested
the tool list (the server exited before answering) · what the tools do when called with real credentials

What this means for the label: Verified here covers what ran. The server's tools were not exercised with a real credential, so nothing about their behavior is asserted; the credential summary below says what to do with that.

Before you give it your token

It will ask for
AI_GATEWAY_API_KEYAI_GATEWAY_TOKENANTHROPIC_API_KEYANTHROPIC_AUTH_TOKENCLOUDFLARE_API_TOKENGEMINI_API_KEYGOOGLE_API_KEYMIN_TOKEN_LENGTHOPENAI_API_KEYOPENAI_COMPATIBLE_API_KEYSLACK_WEBHOOK_URLConfiguration, not credentials: AI_GATEWAY_URL, CLOUDFLARE_ACCOUNT_ID, GATEWAY_PATHS, OPENAI_COMPATIBLE_BASE_URL, OPENAI_COMPATIBLE_MODEL, SMTP_FROM, SMTP_URL
Where it talks to
analitikisler.com (its own)Also cookielaw.org, cookiebot.com, slack.com, deepseek.com, vercel.sh, cloudflare.com, openxmlformats.org, google.com, facebook.com — not the publisher's domain; check what for before connecting.
What the honeypot saw
no credential files read · no network
What we could not test
the tool list (the server exited before answering) · what the tools do when called with real credentials

Connect it with a scoped credential, under the guard, and read the third-party hosts first. The code names cookielaw.org, cookiebot.com, slack.com, deepseek.com, vercel.sh, cloudflare.com, openxmlformats.org, google.com, facebook.com as well as its own vendor. The guard injects AI_GATEWAY_API_KEY, AI_GATEWAY_TOKEN, ANTHROPIC_API_KEY, ANTHROPIC_AUTH_TOKEN, CLOUDFLARE_API_TOKEN, GEMINI_API_KEY, GOOGLE_API_KEY, MIN_TOKEN_LENGTH, OPENAI_API_KEY, OPENAI_COMPATIBLE_API_KEY, SLACK_WEBHOOK_URL only into requests to analitikisler.com (and its subdomains) and refuses the rest.

$vet-skills run --secrets AI_GATEWAY_API_KEY=analitikisler.com,AI_GATEWAY_TOKEN=analitikisler.com,ANTHROPIC_API_KEY=analitikisler.com,ANTHROPIC_AUTH_TOKEN=analitikisler.com,CLOUDFLARE_API_TOKEN=analitikisler.com,GEMINI_API_KEY=analitikisler.com,GOOGLE_API_KEY=analitikisler.com,MIN_TOKEN_LENGTH=analitikisler.com,OPENAI_API_KEY=analitikisler.com,OPENAI_COMPATIBLE_API_KEY=analitikisler.com,SLACK_WEBHOOK_URL=analitikisler.com -- npx -y @analitikisler/ai-tag-debugger

Never paste a credential into a chat, a ticket or a report, including ours. How the guard holds it.

Is [email protected] safe to install?

KillerSkills ran it in the KillerSkills honeypot sandbox (2 scenarios; startup only; authenticated tool behavior not tested) on 2026-11-19 and scored it 81/100 (Verified). In what ran, nothing we could detect went wrong: no undeclared network egress, no credential reads, no persistent instruction writes. What it does with a real credential was not observed; the static findings and the publisher carry the rest of the score. Summary: no network, touches nothing outside the project, writes no memory.

Behavior trace

Honeypot run · 31 s · 2 scenarios · 0 deviations
t+CategoryObserved action
0.0sprocessno install scripts in bundle
10sprocessdependency install exit 0
30sclaimedstartup only: exited (0) before answering · aitd run --site https://example.com --journey purchase aitd "check my GA tags on example.com" (shows the command first; -y skips the question) Run "aitd <command> --help" for that command's options. -h, --help -v, --version AI: anthropic (claude-opus-5-5) · also set: openai (pick with --provider)
30sprocessscript backend runs setup only
30sprocessno install scripts to re-run

Static findings

MEDIUM
File type combined with upload actionsrc/mcp.js:634 · File type combined with upload action — agent may exfiltrate local data files…the verdict, findings, consent and journeys in about 120 lines. summary: summary.json, the counts, journeys and findings as JSON. timeline: one JSON entry per step of each journey (hits with their consent, consent commands, the Google tag'…Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.
AST07
MEDIUM
npx invocation of non-well-known packagesrc/browser.js:28 · npx invocation of non-well-known package — may execute untrusted codechild.on("exit", (code) => (code === 0 ? resolve() : reject(new Error(`Installing the browser failed (exit code ${code}). Run: npx -y [email protected] install chromium`))));Fix: Pin versions and sources; vendor anything fetched at install time.
AST06
MEDIUM
npx invocation of non-well-known packagesrc/doctor.js:35 · npx invocation of non-well-known package — may execute untrusted code…hromium at ${chromiumPath}` : `Playwright ${pw} has no Chromium installed. Run: npx -y playwright@${pw} install chromium` }); // The same line `aitd` with no argum…Fix: Pin versions and sources; vendor anything fetched at install time.
AST06
HIGH
Hidden zero-width characters detectedsrc/report.js:557 · Hidden zero-width characters detected — may conceal contentreturn "" + [header, ...rows.map((r) => r.cells)].map((row) => row.map(csvCell).join(",")).join("\n") + "\n";Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST07
HIGH
Slack webhook URLsrc/notify.js:3 · Slack webhook URL — data will be sent to Slack// "notify": { "slack": "https://hooks.slack.com/services/…", "email": "[email protected]" }Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.
AST07
MEDIUM
File type combined with upload actionsrc/cli.js:368 · File type combined with upload action — agent may exfiltrate local data filesconsole.log(`\nNo plan yet for this site. It already sends ${plan.events.length} ${plan.events.length === 1 ? "event" : "events"}: ${plan.events.map((e) => `${e.platform}:${e.name}`).join(", ")}`);Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.
AST07
MEDIUM
npx invocation of non-well-known packagesrc/mcp.js:486 · npx invocation of non-well-known package — may execute untrusted codereturn fail(connect ? `Could not connect to Chrome: ${err.message}` : `Could not start the browser: ${err.message}. If Chromium is missing, run: npx playwright install chromium`);Fix: Pin versions and sources; vendor anything fetched at install time.
AST06
LOW
Prompt injectionREADME.tr.md:258 · Prompt injection — known jailbreak technique reference · informational: not an instruction file…ktronik tabloyu okuyup yeni bir plan klasörü olarak kaydeder. Google E-Tablolar'dan indirirken Dosya > İndir > Microsoft Excel (.xlsx) seçin ya da her sekmeyi CSV …Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST01
LOW
Authorization escalation languageREADME.md:750 · Authorization escalation language · informational: not an instruction fileThese apps connect to MCP servers over HTTPS (in ChatGPT, through developer mode connectors, depending on your plan). Start the server in HTTP mode and expose it with a tunnel:Fix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
LOW
Prompt injectionREADME.tr.md:60 · Prompt injection — known jailbreak technique reference · informational: not an instruction file…r bilgi bulgusu ekler: konteynerin ad platformlarına ilettiği event'ler tarayıcıdan geçmez, bu yüzden burada görülemez. `gtm_container` dışa aktarımı verildiğinde…Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST01
LOW
Prompt injectionREADME.md:750 · Prompt injection — known jailbreak technique reference · informational: not an instruction fileThese apps connect to MCP servers over HTTPS (in ChatGPT, through developer mode connectors, depending on your plan). Start the server in HTTP mode and expose it with a tunnel:Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST01
LOW
Prompt injectionREADME.tr.md:329 · Prompt injection — known jailbreak technique reference · informational: not an instruction file…e ne kadar sürede yüklendiğini, container'ın okuduğu dataLayer'ı (`l=`), Google'dan mı yoksa sitenin kendi alan adından mı (Google etiket ağ geçidi ya da bir sunuc…Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST01
LOW
Prompt injectionREADME.tr.md:62 · Prompt injection — known jailbreak technique reference · informational: not an instruction file…kte; değer, GTM'in Data Layer Variable'larının okuduğu gibi birikimli dataLayer'dan okunur), bir istisna tetikleyicisi engellemiş ya da event tetiklendiğinde etike…Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST01

Cost to the agent

Instructions loaded
≈93 tok
1 file · 62 in bundle
Tool list per prompt
≈0 tok
0 tools · 0 per tool
Handshake
no reply
launch to tools/list, stdio
Undocumented tools
0
description under 20 chars · avg 0

Tokens estimated at four characters each. The tool list is re-read by the agent on every turn, so it is the number to compare between servers that do the same job: ranked across the index.

Known vulnerabilities

No advisories in OSV for @analitikisler/ai-tag-debugger@0.22.8 as of 4 h ago. This is the package itself, not its dependencies.