Hey there!
Writing this from a hotel hot tub at the coast as I’m here to celebrate my sister's birthday.
It's been an interesting week outside of the day job. I was able to connect with a few people to talk content strategy for the year. It’s always refreshing to hear how others are staying current on everything going on.
I've also been going deep on leveraging AI as a workbench for testing, using Claude Code as my primary driver with Daniel Miessler's PAI layered on top. If you haven't experimented with this stuff yet, you're sleeping on a real edge. More on that in a bit.
As you know, I've spent the past year building out a content pipeline that collects, scores, and surfaces high-signal bug bounty intel from around the world with thousands of pieces sitting in my database. What makes it into this newsletter is only about 5% of what I collect. Based on that data, I've been putting together a Bug Bounty 2026 Resource Guide and I plan on launching it this week. If you see people posting about it on social, a repost or comment would help a ton. Don't worry, I'll make sure a copy goes out here too.
Anyway, let's dive in.
P.S. Reply to this email if you'd want me to do an agentic testing resource drop. It's a topic I'm deep in right now and probably relevant to a lot of you.
I’m available for 1:1 calls if you want to chat about bug bounty, career growth, community building, or anything else you think I can help with. You can book time with me here.
One-Click RCE on Clawdbot/Moltbot via Ethiack’s Autonomous Hacking Agent [Spotlight]
by Ethiack (@Ethiack)
Ethiack shows its autonomous agent discovering exposed Clawdbot/Moltbot deployments and chaining misconfiguration into remote code execution in under two hours. The write-up focuses on real-world operational failurespublicly reachable panels, open ports, and temporary/default settings that were never rolled backand how an agent can pivot from discovery to compromise. The main takeaway is that self-hosted AI assistants with broad integrations and filesystem access create high-impact attack surfaces, and automation compresses time-to-exploit enough that secure defaults and network isolation are mandatory.
Have something you want to Spotlight? Tell me.
HackerOne Launches Agentic PTaaS (AI + Human-Validated Pentesting) [𝕏 Tweet]
by HackerOne
HackerOne announces their Agentic PTaaS product offering.
Bugcrowd Publishes Inside the Mind of a Hacker Report [Blog]
by Bugcrowd
Bugcrowd released its Inside the Mind of a Hacker report based on survey data from ~1,300 researchers. It covers researcher demographics, motivations, AI usage patterns, and an increased focus on hardware security, making it more useful for program strategy than exploitation technique.
NahamSec Hosts Snyk Fetch The Flag [𝕏 Tweet]
by Ben Sadeghipour (@NahamSec)
NahamSec announces an upcoming Snyk Fetch The Flag event and links to signup details. Its a community/event update rather than a technical post.
Bug Bounty Masterclass: 3.5-Hour YouTube Course [𝕏 Tweet]
by Gal Nagli (@galnagli)
This tweet shares a 3h39m Bug Bounty Masterclass published on YouTube as a single-session, end-to-end course. Its positioned as a long-form training resource in the LLM deep-dive style rather than a specific technique write-up.
Did I miss an important update? Tell me.
Caido Plugins: NTLM Authentication + Tor Routing [𝕏 Tweet]
by TheSytten
This tweet highlights two newly released Caido plugins: one adding NTLM authentication support and another routing traffic via Tor. Its a practical quality-of-life update for testing NTLM-protected services and controlling egress during assessments.
Burp AI Agent v0.1.1 Ships OpenAI-Compatible Backend + Targeted Tests [𝕏 Tweet]
by Six2dez (@Six2dez1)
Six2dez announces burp-ai-agent v0.1.1, adding an OpenAI-compatible backend (to support multiple providers), a targeted tests submenu, and improved support for custom headers/API keys with Ollama and LM Studio. The update also mentions persistent chat context and general fixes/QoL improvements.
Hurricane Electric Cert Search (Certificate Transparency + TLS Tester) [Blog]
by Hurricane Electric
Hurricane Electrics Cert Search provides a CT-backed certificate search with subdomain inclusion and quick navigation into certificate details. The site also includes a web-based SSL/TLS tester for HTTPS and common STARTTLS services, useful for recon and fast configuration validation.
PentestAgent is an open-source AI agent framework aimed at black-box security testing workflows across bug bounty, red teaming, and pentesting. It ships a modular adapter architecture, persistent workspace/RAG context, CLI/TUI interfaces, Docker deployment, and integrations like Metasploit and Playwright. The project is built around orchestrating recon-to-exploitation tasks while keeping the system extensible via adapters.
Have a favorite tool? Tell me.
Hacking an AI Children's Toy: Remote Access to Every Conversation [Blog]
by Joseph Thacker (@rez0)
Joseph Thacker analyzes an LLM-enabled childrens toy and reports backend exposure that allowed access to conversation data. The investigation started from a console subdomain referenced in CSP headers and led to an admin interface reachable via a Google OAuth flow with insufficient authorization gating. Its a strong reminder that conversational/IoT telemetry should be treated as highly sensitive and that admin consoles must be isolated and properly access-controlled.
PackageGate: Six JS Package-Manager Bugs That Break Common Supply-Chain Mitigations [Blog]
by Oren Yomtov (@orenyomtov)
This post documents PackageGate : six vulnerabilities across JavaScript package managers (including pnpm, vlt, and Bun) that bypass widely recommended defenses like disabling lifecycle scripts and relying on committed lockfiles. It includes CVEs for pnpm, notes patch status across vendors, and discusses npms reported lack of action. The main point is that standard mitigations can be bypassed in practice, so supply-chain assumptions need to be revalidated per toolchain.
Parse and Parse: MIME Validation Bypass to XSS via Parser Differential [Blog]
by Critical Thinking Podcast
This research note demonstrates a Content-Type parser differential where a crafted value can be interpreted differently by server-side MIME parsing libraries and browsers, enabling reflected XSS. It walks through how header coalescing and last-match behavior in Chromium/Firefox can diverge from common backend parsing, and why using the raw, unnormalized header value is dangerous. The takeaway is to validate and consume the parser-normalized Content-Type rather than the original string.
SmarterMail WT-2026-0001: Auth Bypass Enabling Admin Password Reset [Blog]
by watchTowr (@watchtowrcyber)
watchTowr Labs details WT-2026-0001, an authentication bypass in SmarterTools SmarterMail that allows resetting the system administrator password. The post notes this can lead to OS command execution via SmarterMails existing administrative functionality and references reports of in-the-wild exploitation. SmarterTools issued an emergency patch in release 9511.
Instagram Bug Exposed Private Posts via Server-Side Authorization Bypass [Blog]
by Jatin Banga
This write-up describes an authorization failure that returned private Instagram post data to unauthenticated requests when specific mobile headers were used, with the data embedded in the HTML response. It includes a PoC script, logs, and a timeline, and notes the exposure was conditional (only some private accounts leaked), complicating reproducibility. The post also highlights how silent mitigations and unreproducible closures can create friction in disclosure.
Account Takeover via HTML Injection (Write-up + PoC) [𝕏 Tweet]
by zack0x01
This tweet links to a detailed write-up showing an account takeover chain starting from HTML injection, with a full PoC. Its a useful reminder that not quite XSS injection classes can still become high-impact depending on where the HTML lands and what tokens/actions are reachable.
Hono IP Restriction Middleware: IPv4 Validation Bypass Enables IP Spoofing (CVE-2026-24398) [Write-Up]
by honojs/hono
This advisory covers CVE-2026-24398: an IPv4 parsing/validation flaw in Honos IP Restriction Middleware that allows octets >255 to overflow and be reinterpreted as different IPs. In deployments that trust client-supplied IP headers (for example X-Forwarded-For), this can enable allowlist/blocklist bypass via spoofed addresses. The issue is fixed in 4.11.7.
Did I miss something? Tell me.
postMessage Vulnerabilities: Detection and Exploitation Techniques [𝕏 Tweet]
by Intigriti
Intigriti shares an article on finding and exploiting postMessage flaws, including origins validation issues that can lead to DOM XSS and sensitive data exposure. The focus is on practical detection patterns in cross-document messaging handlers and common implementation mistakes.
AI Red Teaming Is Not Red Teaming (But It Kinda Is) [Blog]
by Joey Melo (@MrJoeyMelo)
This piece contrasts traditional red teaming with AI red teaming, arguing the mechanics differ while the adversarial mindset stays similar. It frames prompt-based manipulation as a form of social engineering against model behavior and guardrails, and discusses how objectives and constraints shift when the target is a model rather than defenders. Its primarily conceptual rather than technique-driven.
Intigritis 31 Bite-Sized Bug Bounty Tips and Resources for 2026 [Blog]
by Intigriti
This Intigriti roundup compiles 31 short tips and links spanning recon, tooling, reporting, and productivity. Its structured as a checklist-style refresher rather than original research, and works well as onboarding material or a quick workflow audit.
HackerOne Interview: Evan Connelly on Bug Bounty and Non-Linear Career Paths [𝕏 Tweet]
by Evan Connelly (@evanconnelly)
This tweet shares a HackerOne interview with Evan Connelly, covering his background and how he approaches security research alongside work outside of traditional security roles. Its a career/community perspective rather than a technical deep dive.
Did I miss something? Tell me.
Lessons from $1M+ in Three Years of Bug Bounties [Video]
by Ben Sadeghipour (@NahamSec)
This tweet links to a video where NahamSec reflects on the habits, process, and decision-making that led to $1M+ in bounties over three years. Its a strategy-and-workflow retrospective rather than a technical walkthrough.
Methodology for Turning XSS into Account Takeover [Video]
by Harrison Richardson (@rs0n)
This video outlines an exploit methodology for escalating from XSS into account takeover, focusing on practical chaining rather than standalone payloads. It discusses browser-side primitives used to hijack sessions or perform authenticated actions and highlights common implementation pitfalls that make XSS materially exploitable.
AI Just Made Bug Bounty Way Easier [Video]
by Amr Elsagaei (@amrelsagaei)
This video demos an AI-assisted workflow for testing source-available/OSS-style bounty targets using repo-aware tools to rapidly navigate code and identify likely vulnerable paths. It walks through deploying a target, interrogating code with vulnerability-focused prompts, and validating results with manual reproduction. The emphasis is on speed gains from AI triage while keeping human verification in the loop.
How Hackers Analyze GraphQL Responses for High-Impact Bugs? [Video]
by Medusa (@medusa_0xf)
This video covers practical GraphQL recon using introspection to enumerate types, queries, and sensitive fields. It shows how to read introspection responses, identify high-value operations, and test for authorization gaps that can lead to data exposure or privilege escalation. The core theme is turning schema visibility into targeted, high-signal test cases.
Fuzzing and Scanning Workflows for Bug Bounty [Video]
by coffinxp7
This video discusses combining scanners and fuzzers with tuning strategies to reduce noise and improve hit quality during bounty testing. It covers wordlists, rate limiting, scope awareness, and practical prioritization to turn automated results into actionable leads. The emphasis is on operational tradeoffs rather than a single vulnerability class.
Did I miss something? Tell me.
GraphQL Access-Control Bypass via Escaped Field Name [𝕏 Tweet]
by the_IDORminator (@the_idorminator)
This tweet describes a GraphQL authorization bypass where an escaped quote in a field name caused the backend to skip an access-control check and return PII for arbitrary identifiers. Its a good example of parser edge-cases turning into IDOR-like data exposure.
XSS Cheat Sheet Updated to 1,337 Test Vectors [𝕏 Tweet]
by Gareth Heyes (@garethheyes)
Gareth Heyes shares an update to the XSS Cheat Sheet, expanding it to 1,337 test vectors. Its a dense payload/reference set for validation and edge-case coverage when probing filters and contexts.
Find Exposed Hosts via Shodan and FOFA [𝕏 Tweet]
by Behi
This tweet points to Uncover, a recon utility from pdiscovery that identifies exposed hosts by querying internet search engines like Shodan and FOFA. Its geared toward accelerating asset discovery and expanding target surface area.
Facebook Mobile App Testing: Intercepting Traffic via Native Library Patching [𝕏 Tweet]
by Andre9 Baptista (@0xacb)
Andre9 Baptista reflects on prior success intercepting Facebook mobile traffic by patching a native library, suggesting the technique remains relevant for newer app features. Its more of a directional nudge than a full methodology, but it highlights where mobile testing effort can still pay off.
Single-Quote Smoke Tests Still Catch Real SQLi [𝕏 Tweet]
by the_IDORminator (@the_idorminator)
This tweet argues that quick single-quote smoke tests still reliably surface SQL injection, especially in exposed web services discovered via JavaScript. It also notes a responsible workflow of validating impact without over-exploiting once injection is confirmed.
H1-702 LHE 2018: Unattended Laptop Takeover [𝕏 Tweet]
by Yassine Aboukir (@Yassineaboukir)
Yassine Aboukir resurfaces a HackerOne report from H1-702 LHE 2018 describing a takeover of an unattended work laptop. Its a useful reminder that live events add physical and social-engineering risk that doesnt show up in typical web scopes.
Intigriti: 5 Practical Tips for New Bug Bounty Hunters [𝕏 Tweet]
by Intigriti
Intigriti shares five beginner-focused tips: spend meaningful time per target, prioritize reading JavaScript over blind scanning, treat duplicates as leads, use docs/changelogs to expand attack surface, and stay consistent. Its lightweight guidance, but the emphasis on depth over breadth is directionally correct.
Did I miss something? Tell me.
Because Disclosure Matters: This newsletter was produced with the assistance of AI. While I strive for accuracy and quality, not all content has been independently vetted or fact-checked. Please allow for a reasonable margin of error. The views expressed are my own and do not reflect those of my employer.