LLM, mobile, pollution — then chain, triage, get paid.

0:00–0:15 OWASP LLM shapes: direct/indirect/RAG/agent perms

Define 4 + risk each. Read OWASP LLM Top 10 2025 (LLM01 injection + LLM08 vector/RAG) summaries + Agentic 2026 ASI01 tool-abuse/excessive-agency summaries ONLY.

0:15–1:05 PortSwigger Web LLM labs: prompt → tool → exfil (lab)

Do PortSwigger LLM labs (excessive-agency + indirect injection + insecure output handling) IN ORDER (lab). Chain: prompt → tool call → output → OOB. YT: PortSwigger LLM attacks indirect prompt injection lab.

1:05–2:00 Tool-output exfil OOB + RAG-poison persist + ASCII/tag smuggle

Test image-markdown ![x](http://CB/x) OOB + ASCII/Unicode tag smuggle + system-prompt leak probe + IDOR-via-AI (cross-tenant doc) + RAG doc-poison persist (lab, own corpus). Read OWASP agentic threats § tool-output/RAG rows ONLY. X: LLM prompt injection bounty.

2:00–2:35 Prompt matrix: direct/indirect/tenant/RAG/agent + perms

Build matrix + agent-permission audit (read vs write vs exec tools). Note tenant isolation (IDOR-via-AI).

2:35–3:00 Direct-vs-indirect test + FP gate + gaps

Distinguish + apply FP gate to 1 candidate (kill or keep with proof). Triage gaps.

0:00–0:15 Mobile gaps: pinning/storage/deeplink/version-lag

List 4 + test idea each. Read OWASP MASVS storage/network/auth headings + HackTricks Android pentesting § storage/deeplink ONLY.

0:15–1:05 Pinning-bypass theory + token-storage + deeplink (lab)

Study objection -g com.app explore --startup-command 'android sslpinning disable' + frida -U -f com.app -l pinning.js patterns (lab device ONLY) + THM androidhacking101 Task 2 (APKTool/Jadx static) + Task 4 (Frida) ONLY. Review token storage (SharedPrefs/Keychain) + app:// params (lab app).

1:05–2:00 Grab deeplink leak + pinning-bypass IDOR

Read 1 Grab deeplink sensitive-leak (token in app:// logged) + 1 pinning-bypass→IDOR (mobile misses web fix) — leaked field + version-gap ONLY each.

2:00–2:35 Mobile-vs-web version diff map

Map mobile API vs web API (version/authz delta). Note web-fixed-but-mobile-open candidates.

2:35–3:00 Deeplink-risk test + gaps

Explain token-via-deeplink risk + log exposure. Triage gaps.

0:00–0:15 __proto__/constructor/gadget idea

Write ?__proto__[x]=polluted + constructor.prototype + gadget (XSS/RCE sink). Read PortSwigger pollution topic § pollute→gadget ONLY.

0:15–1:05 Pollute → gadget → XSS (lab)

Do PortSwigger pollution labs (client-side pollute→DOM XSS + browser-API gadget) + THM xss Task 5 revisit (DOM) ONLY. Test __proto__ then gadget trigger (lab). YT: prototype pollution to XSS exploit 2025.

1:05–2:00 Pollution→XSS chain + gadget-RCE note

Read 1 pollution→XSS (query→proto→sink) + 1 Node gadget-RCE note (server-side impact) — polluted key + gadget ONLY each. X: prototype pollution XSS bounty.

2:00–2:35 Gadget map + server-impact note

Map client gadgets (DOM) + server gadgets (Node). Note CSP/probe bypass via gadget.

2:35–3:00 Gadget-idea test + gaps

Explain your best gadget end-to-end. Triage gaps (DOM/misc next).

0:00–0:15 DOM sinks + misc patterns + brute math

List sinks (clobber/postMessage/SW) + misc (soft-delete/invite/SAML/ReDoS) + OTP 10^6 math. Read PortSwigger DOM § sinks + AnukarOP index headings ONLY.

0:15–1:05 Clobber/postMessage/SW + SAML replay drills (lab)

Drill DOM clobber + postMessage origin + service-worker abuse + SAML replay (SAMLRaider revisit) on LAB ONLY per HackTricks postMessage § + PortSwigger DOM-clobber lab ONLY.

1:05–2:00 DOM-clobber ATO + invite-bypass priv-esc

Read 1 DOM-clobber ATO + 1 invite-bypass (token reuse → admin) — clobbered node / reused token ONLY each. X: DOM clobbering ATO + invite bypass privilege escalation.

2:00–2:35 Soft-delete/invite/SAML/ReDoS sweep (lab + gentle VDP reads)

Sweep deleted-object access + invite reuse + SAML replay + ReDoS regex timing (single request, baseline). PATT Account Takeover + SecLists 10-line brute-math demo (lab). H1 rate limit OTP search for brute-gate lesson.

2:35–3:00 Misc-triage test + gaps

Judge 3 misc candidates (pay/N/A + reason). Triage gaps (chaining next).

0:00–0:15 3 chain shapes: XSS→ATO, SSRF→cloud, self-XSS→CSRF

Diagram handoffs (cookie/session, role-creds/s3-ls, frameless-CSRF→stored). Read HuntBook Phase-4 chains headings ONLY.

0:15–1:05 Build 3 chains on labs (minimal PoCs)

Build XSS→session-theft→account-action + SSRF→role→s3-ls + self-XSS→CSRF→stored (lab, minimal requests). Trim each to fewest steps. YT: bug bounty chaining low to critical methodology. X: chaining XSS ATO writeup.

1:05–2:00 XSS→ATO + SSRF→cloud full chains (handoff study)

Read PATT Account Takeover 1 XSS→ATO + 1 SSRF→cloud — handoff request + impact delta ONLY each. Note CVSS jump (low→high/critical).

2:00–2:35 Severity-escalation worksheet per chain

Score before/after per chain via first.org calculator base ONLY. Note which link in chain carries the severity.

2:35–3:00 Chain-value test + gaps

Explain why each chain pays more than parts. Triage gaps (triage/report next).

0:00–0:15 Kill-fast gates + CVSS base metrics

Write gates (repro? impact? scope? dup? minimal proof?) + base metrics (AV/AC/PR/UI/S/CIA). Read HuntBook triage-gate headings + first.org metric defs ONLY.

0:15–1:05 Gate 3 old candidates + kill reasons

Pull 3 prior candidates. Gate each: keep (report) or kill (reason: self-only/dup/no-impact/unrepro). Log kills with one line each.

1:05–2:00 Accepted vs N/A pair + duplicate-avoidance drill

Read H1 hacktivity 1 accepted + 1 N/A same class (takeover/XSS/IDOR) — decision reason ONLY each + dup-check: same-program hacktivity + pentester.land + bugpedia before drafting (3 searches, screenshot queries). X: @corb3nik @rez0__ for triage threads.

2:00–2:35 2 full reports: repro + impact + remediation + CVSS

Write 2 reports in AnukarOP playbook structure (repro steps + request + impact + CVSS + fix) + PATT business-logic/account-takeover remediation quotes ONLY. Evidence hygiene: redacted transcripts, URL+timestamp shots, no cookie/PII.

2:35–3:00 Kill-rules test + gaps

Recite gates. Assign report backlog to Day 7 hunt.

0:00–0:15 Hunt order: authz → fetch → logic, newest first

Write order + changelog/new-endpoint priority rule. Pick 1 VDP + 1 wide-scope paid (H1 directory filter: cloud/internet, wildcard) for today.

0:15–1:05 Compressed hunt: recon→pollution/DOM/RAG checklist (log everything)

Run authz→fetch→logic + Day-1 RAG + Day-3 pollution quick-pass on newest endpoints (gentle, in-scope). Log every request (history on). TBHM recon bullets + JHaddix mapping order as checklist.

1:05–2:00 Same-program disclosed patterns (hunt with answers)

Read H1 hacktivity same-program (pollution/DOM/LLM/OTP) 2 — pattern ONLY. Re-test those patterns first on newest code.

2:00–2:35 Reports sent + 30-60-90 + private-invite track

Send/finalize reports (Day-6 template). Write 30d (triage/report muscle, VDPs) → 60d (chaining, 1 paid) → 90d (paid hunt, invite track via quality VDP reports). Note rotation hygiene (keys/certs redacted, sessions rotated).

2:35–3:00 Where-bugs-live test + program complete

Answer: newest-30-days + least-tested surface + your next 3 targets. Confirm all 12 weeks ticked (index → each week %).