[ top 10 most noteworthy of 113 repo events in this month // 30d ]
Scanner authors dominated the stretch: Assetnote shipped three high-fidelity detectors for the php2shell and node2shell waves, xssdoctor kept pushing inert Git-client execution harnesses, and vitorfhc's videt formalized the negative-day patch-diffing game just as the N-able zero-day proved the commit-to-advisory window is now measured in hours. Between the 62-level-deep PHP array probes and the loopback protocol-v2 fixtures, the repos this month read like a field manual for confirming -- not just claiming -- exploitability.
assetnote/php2shell-scanner -- Python scanner for CVE-2026-109618/109619 that sends multipart control requests and probes with 62-level-deep PHP array keys scaling from 1MiB to 20MiB. High-fidelity detection without firing blind payloads at prod; the key-depth scaling is the interesting part to read.
assetnote/node2shell-scanner -- probes Brotli/JSON content-type handling in Node.js HTTP apps, then streams a ~23MB payload to confirm CVE-2026-111497 exploitability. Pairs with the Wiz writeup; the content-type gate logic is worth diffing against your own detection scripts.
xssdoctor/git-cve-2026-113408-poc -- loopback-only smart-HTTP fixture proving git clients accept an unsafe unborn HEAD symref from a protocol-v2 server, with branch-update resolution escaping the .git directory. Clean, inert oracle you can run locally in five minutes.
vitorfhc/videt -- watches major OSS repos for silent security patches landing before a CVE is assigned, then checks whether the fix is actually reachable/exploitable. Formalizes the negative-day hunt that this month's N-able and gVisor silent-cleanup commits both point at.
dawgyg/AppleResearch -- raw C dump of macOS vulnerability research the author believes should be patched but went unreported for missing bounty thresholds. 510 stars in days; a large corpus of unreported findings, some plausibly portable to iOS and visionOS.
Python scanner giving high-fidelity detection of CVE-2026-111497 ("node2shell") in Node.js HTTP apps, probing Brotli/JSON content-type handling then streaming a ~23MB payload to confirm exploitability.
google/security-research is the advisory + PoC drop for bugs found in non-Google code; new commits land a CVE-2026-98164 kvmCTF host-crash exploit plus kernelCTF writeups for CVE-2026-80521_lts and CVE-2025-40214_lts_cos.
A new Python PoC harness for CVE-2026-113408, using a loopback-only smart HTTP fixture to test how a Git client accepts an unsafe unborn HEAD symref from a protocol-v2 server. It demonstrates branch-update resolution escaping the .git directory.
Python scanner for high-fidelity detection of php2shell (CVE-2026-109618/109619), sending multipart control requests and probes with 62-level-deep PHP array keys scaling from 1MiB to 20MiB.
AppleResearch is a raw C-language dump of the author's macOS vulnerability research, much of which he believes should be patched but went unreported for not meeting bounty thresholds. It offers a large corpus of macOS bug-hunting findings, some potentially portable to other Apple OSes.
CSPBypass.com is a tool helping ethical hackers bypass restrictive Content Security Policies and exploit XSS on sites whose CSPs only whitelist certain domains. The latest update fixes the CSP matcher to handle scheme-relative and multiple directives.
Query-param PoC server: spins up shareable HTML/JS exploit pages with custom response headers baked into the URL, handy for header-gated XSS proof. Latest commits add nested autocomplete in SVG/XML plus a scrolling fix on small screens.
videt: watches major OSS repos for silent security patches landing before a CVE is assigned, then checks whether the fix is actually reachable/exploitable. Targets the negative-day window between commit and advisory.
intent-anchor is a Python proof-of-concept harness for long-horizon LLM agents that makes goal adherence mechanical by storing hard constraints in an immutable spec instead of trusting the prompt. It aims to prevent agent drift on extended tasks.
An inert local partial-repository fixture for authorized testing of Git's partial-clone behavior, tied to CVE-2024-32004. Its config contains a promisor upload-pack command that only writes a fixed marker, confirming code execution.
source: Assetnote Official (@assetnote) | role: Guest Org (Shubs & Sean Yeoh) | published: 2026-10-25 04:24 UTC
[ notes ] Python scanner giving high-fidelity detection of CVE-2026-111497 ("node2shell") in Node.js HTTP apps, probing Brotli/JSON content-type handling then streaming a ~23MB payload to confirm exploitability.
[ source excerpt ] Repo description: High fidelity detection mechanism for node2shell (CVE-2026-111497). README: # node2shell-scanner A command-line tool for detecting CVE-2026-111497 in Node.js HTTP applications. ## How It Works The scanner sends four one-byte controls to learn how the endpoint handles valid Brotli JSON, malformed Brotli, requests without a JSON content type, and unsupported encodings. It then streams a 23,068,764-byte Brotli probe and repeats the valid control. A target is reported vulnerable only when the full probe reaches the valid route-response class, the malformed-response class is distinct, and the final control confirms the endpoint remained stable. This matches Assetnote's internal postmodule detection logic and does not exec...
source: Michael Cote & Darby Hopkins (@google) | role: Guests (Pinned Security Repo) | published: 2026-11-13 13:31 UTC
[ notes ] google/security-research is the advisory + PoC drop for bugs found in non-Google code; new commits land a CVE-2026-98164 kvmCTF host-crash exploit plus kernelCTF writeups for CVE-2026-80521_lts and CVE-2025-40214_lts_cos.
[ source excerpt ] About google/security-research: This project hosts security advisories and their accompanying proof-of-concepts related to research conducted at Google which impact non-Google owned code.. Recent commits: Add CVE-2026-98164 kvmCTF host-crash exploit (#440) | kernelCTF: add CVE-2026-80521_lts (#431) | Add kernelCTF CVE-2025-40214_lts_cos (#347)
source: Jonathan Dunn (@xssdoctor) | role: Co-Host & Recurring Guest | published: 2026-11-13 12:46 UTC
[ notes ] A new Python PoC harness for CVE-2026-113408, using a loopback-only smart HTTP fixture to test how a Git client accepts an unsafe unborn HEAD symref from a protocol-v2 server. It demonstrates branch-update resolution escaping the .git directory.
[ source excerpt ] Newly created repository.README: # CVE-2026-113408 bounded Git PoC This repository contains a loopback-only smart HTTP fixture for testing CVE-2026-113408: a Git client can accept an unsafe unborn `HEAD` symref from a protocol-v2 server and later resolve the first branch update outside `.git`. The harness creates one empty bare repository and changes only the unborn HEAD target in the server's `ls-refs` response: ```text refs/heads/../../../../CVE_2026_113408_MARKER.txt ``` The first commit in a vulnerable clone can therefore write the new commit ID to the fixed marker in the clone's dedicated parent. ## Important GitHub limitation GitHub hosts this source and reproduction h...
source: Assetnote Official (@assetnote) | role: Guest Org (Shubs & Sean Yeoh) | published: 2026-10-24 03:40 UTC
[ notes ] Python scanner for high-fidelity detection of php2shell (CVE-2026-109618/109619), sending multipart control requests and probes with 62-level-deep PHP array keys scaling from 1MiB to 20MiB.
[ source excerpt ] Repo description: High fidelity scanner for php2shell (CVE-2026-109618 and CVE-2026-109619). README: # php2shell-scanner A command-line tool for detecting CVE-2026-109618 and CVE-2026-109619 in PHP applications. ## How It Works The scanner sends a minimal multipart control request followed by up to three crafted multipart probes. The probes use PHP array keys nested 62 levels deep and increase from 1 MiB to 20 MiB, matching the detection logic used by Assetnote's internal scanner. A target is reported vulnerable when the probe produces a PHP memory-exhaustion error, a differential empty HTTP 500, a PHP-FPM worker failure, or an advertised PHP version known to be unpatched. Initial 5xx responses, common request-rejection statuses, and nonempt...
source: Tommy DeVoss (@dawgyg) | role: Guest | published: 2026-11-08 04:42 UTC
[ notes ] AppleResearch is a raw C-language dump of the author's macOS vulnerability research, much of which he believes should be patched but went unreported for not meeting bounty thresholds. It offers a large corpus of macOS bug-hunting findings, some potentially portable to other Apple OSes.
[ source excerpt ] Repo description: A raw dump of my Apple MacOS research. Some likely can be ported over to other Apple OSes. Alot of this should likely be patched, but I did not report many of these as they didnt meet the bar for a bounty.. README: # AppleResearch A raw dump of my Apple MacOS research. Some likely can be ported over to other Apple OSes. Alot of this should likely be patched, but I did not report many of these as they didnt meet the bar for a bounty. # Disclaimer: Everything shared here is for educational purposes to help others that are starting out doing legitimate security research against MacOS. Any misuse of this information is not the fault of the author.
source: Rene de Sain (@renniepak) | role: Guest | published: 2026-11-12 15:01 UTC
[ notes ] CSPBypass.com is a tool helping ethical hackers bypass restrictive Content Security Policies and exploit XSS on sites whose CSPs only whitelist certain domains. The latest update fixes the CSP matcher to handle scheme-relative and multiple directives.
[ source excerpt ] About renniepak/CSPBypass: CSPBypass.com, a tool designed to help ethical hackers bypass restrictive Content Security Policies (CSP) and exploit XSS (Cross-Site Scripting) vulnerabilities on sites where injections are blocked by CSPs that only allow certain whitelisted domains.. Recent commits: Merge pull request #53 from edoardottt/main | replace spaces with tabs | fix CSP matcher to handle scheme-relative and multiple src URLs
[ notes ] Query-param PoC server: spins up shareable HTML/JS exploit pages with custom response headers baked into the URL, handy for header-gated XSS proof. Latest commits add nested autocomplete in SVG/XML plus a scrolling fix on small screens.
[ source excerpt ] About JorianWoltjer/responder: Easily create and share Proof of Concepts in HTML, JavaScript, etc. with custom headers, all via query parameters. Recent commits: Support nested autocomplete in SVG/XML | Fix autocomplete error & add langauge shortcut | Fix scrolling on smaller screens
[ notes ] videt: watches major OSS repos for silent security patches landing before a CVE is assigned, then checks whether the fix is actually reachable/exploitable. Targets the negative-day window between commit and advisory.
[ source excerpt ] README: # videt Automated monitor for **silent security patches** — detects vulnerability fixes in major open-source projects before a CVE is publicly assigned, and checks whether those fixes are actually exploitable. ## What problem does this solve? There's a dangerous window between when a security patch lands in source code and when it gets a CVE number. During this "negative-day" period, the vulnerability is effectively public (anyone can read the diff) but most defenders haven't acted yet. videt closes that gap by alerting you the moment a suspicious commit is detected — and telling you whether the fix is complete or still bypassable. ## How it .... Recent commits: Update vulnerability monitor state | Update vulnerability monitor state | Update vulnerability monitor state
source: Nick Copi (@7urb0) | role: Guest | published: 2026-11-04 07:44 UTC
[ notes ] intent-anchor is a Python proof-of-concept harness for long-horizon LLM agents that makes goal adherence mechanical by storing hard constraints in an immutable spec instead of trusting the prompt. It aims to prevent agent drift on extended tasks.
[ source excerpt ] Repo description: Goal-state as a first-class citizen: mechanical intent anchoring for long-horizon agents (POC). README: # intent-anchor > 把「目标状态」从 prompt 里的一段文字，变成执行中强制维持的一等公民结构。 **English summary**: `intent-anchor` is a proof-of-concept harness for long-horizon LLM agents. Instead of trusting the model to "stay on task", it makes goal adherence **mechanical**: hard constraints live in an immutable spec (not a system prompt), every tool call is forced through a gateway that re-injects goal state, deviations require evidence-backed approval by a deterministic arbiter (fail-closed), and downstream agents must explicitly accept or challenge upstream deviations across multi-hop chains — silent inheritance is impossible. Live-verified with DeepSeek (OpenAI-compatib...
source: Jonathan Dunn (@xssdoctor) | role: Co-Host & Recurring Guest | published: 2026-11-13 03:20 UTC
[ notes ] An inert local partial-repository fixture for authorized testing of Git's partial-clone behavior, tied to CVE-2024-32004. Its config contains a promisor upload-pack command that only writes a fixed marker, confirming code execution.
[ source excerpt ] Repo description: Inert local partial-repository archive for authorized Muse Git testing. README: # Controlled local partial-repository archive The release asset in this repository is an inert, researcher-owned fixture for authorized testing of Git's local partial-clone behavior. Its local Git config contains one promisor upload-pack command that writes only the fixed text `MUSE_GIT_CVE_2024_32004_FIRED` into its own `.git` directory and exits. It performs no reads, network callbacks, persistence, or writes outside the extracted fixture directory.