Works with the AI Agents you already use
Telegram Mini App & Bot Security Audit
Telegram Mini Apps are everywhere now: shops, games, tap-to-earn, wallets, course platforms.
Free
Telegram Mini App & Bot Security Audit
Example session with this skill installed
security audit this telegram mini app
- Read your context and instructions
- Compiled the telegram mini app
Summary
Real triage output from the bundled scanner on the intentionally vulnerable test project in the repo (tests/fixtures/vuln_app: an aiogram bot, an Express backend and a Mini App frontend). The agent then validates each lead against the code and writes the final report with fixes.
Scanner leads (excerpt)
telegram-miniapp-security-audit · vuln_app · 5 files · 19 rules
CRITICAL TMA-SEC-002 Bot token exposed to the Mini App frontend
webapp/src/main.js:13
HIGH TMA-AUTH-004 Login Widget key derivation used for Mini App initData
server/auth_legacy.py:15
secret = hashlib.sha256(BOT_TOKEN.encode()).digest()
HIGH TMA-SEC-003 Telegram Bot API called from the frontend
webapp/src/main.js:13
HIGH TMA-PAY-001 Purchase granted from the client invoice callback
webapp/src/main.js:27
if (status === 'paid') unlockPremium();
MEDIUM TMA-BOT-001 Admin check by username
bot/main.py:18
MEDIUM TMA-NET-001 Webhook set without a secret token
bot/main.py:28
await bot.set_webhook(WEBHOOK_URL)
MEDIUM TMA-AUTH-003 initData validated without an auth_date freshness check
server/index.js:15
MEDIUM TMA-AUTH-002 Backend trusts a user id from the request
server/index.js:22
const userId = req.body.user_id;
MEDIUM TMA-AUTH-001 Identity taken from initDataUnsafe
webapp/src/main.js:3
MEDIUM TMA-WEB-001 HTML injection sink in the Mini App
webapp/src/main.js:5
LOW TMA-PAY-002 Pre-checkout query approved without checks
bot/main.py:24
...
19 leads (critical: 1 high: 3 medium: 7 low: 7 info: 1)
What happens next
Leads are hints, not findings. The skill has the agent map how identity is established on every route, prove each lead is reachable, compare the initData validator with the bundled reference implementation of Telegram's algorithm, drop false positives, and write REPORT.md plus findings.json with a code-level fix for every confirmed issue.
Connects securely to your tools. The creator never sees your data.
About this skill
Telegram Mini Apps are everywhere now: shops, games, tap-to-earn, wallets, course platforms. Most are built fast, and they break in the same handful of ways: the backend trusts initDataUnsafe.user.id or a user_id in the body, initData is checked with the Login Widget algorithm, the bot token sits in a VITE_ env var, the webhook has no secret_token, or premium unlocks when the openInvoice callback says paid.
This skill gives your agent a Telegram-specific playbook and makes it prove each finding before reporting it.
What it checks
- initData & auth: the official HMAC algorithm,
auth_dateexpiry, identity frominitDataUnsafe, IDOR via client-sent ids - Bot: token in git or the frontend bundle, webhook
secret_token, admins by username, HTML injection in messages - Payments: Telegram Stars and providers,
pre_checkout_querychecks, granting only onsuccessful_payment, idempotent grants - Mini App frontend: XSS,
start_param,openLinkopen redirects, tokens in localStorage, CSP - Backend: IDOR, referral and tap-to-earn abuse, CORS, initData in logs
How it works
- Recon: where identity is established on every route, where value moves.
- A fast, zero-dependency scanner (Node 18+, 19 rules, JSON and SARIF output) collects leads.
- Area-by-area hunting with checklists, then a fresh verifier tries to disprove each candidate.
- You get
REPORT.mdandfindings.jsonwith severity, CWE, proof and a fix in your stack (aiogram, python-telegram-bot, Telegraf, grammY, Express, FastAPI, Laravel, Go, React/Vue, Flutter Web).
Includes a dependency-free reference implementation of Telegram's initData validation with a self-test (rejects forged users, expired data, the wrong token and the Login Widget key).
Open source
Created and maintained by the author. The free, MIT-licensed source, GitHub Action and issue tracker live at https://github.com/MBotirjonDev/telegram-miniapp-audit
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Security.
Requires Node.js for automated scanning. Cannot audit external APIs or database rules missing from the project files. Does not perform live bot API calls.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
1 use
Times used to date
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 19 days ago
- Free with an Agensi account
- Software
Node.js 18+
Python
Git
Node.js runs the bundled scanner; no npm install needed.