phishing_kits - Archive of phishing kits captured from suspicious

项目说明

phishing_kits

Archive of phishing kits collected from suspicious domains detected by
phishunt.io. Over 1,000 archives, collected since 2020
but not in every month (see Co over time). The
exact count is the number of rows in index.csv.

Layout: YYYY/YYYYMM/<host> (<filename>.ext)/<filename>.ext.

Read this before cloning

These are li kits, stored exactly as they were captured: **not
encrypted, not password-protected, not defanged**. That has consequences:

Analyse in an isolated en.

Co over time

Collection has not been continuous. A month with nothing collected has no
directory in the tree and no rows in index.csv, so months are skipped rather
than shown as empty. Rows per month therefore measure what was captured, not
how much phishing existed.

One gap has a known cause. There are no archives for July 2025 through April
2026, ten consecuti, because nothing was being collected: this
repository received no commits at all between 13 June 2025 and 7 May 2026, and
the first capture after that landed on 9 May 2026. git log shows both.

Other months are absent as well. Their cause is not established and is not
guessed at here. also by more than an order of magnitude between
the months that were collected. Treat this as an opportunistic sample rather
than a time series.

Which months exist, and how many archives each holds:

index.csv

A manifest of every archive in the repository.

| column | meaning |
| --- | --- |
| path | repo-relative path of the archive |
| size | size in bytes |
| git_sha1 | git object id, so git hash-object <file> a local copy against this file alone |
| sha256 | file hash, which is what , MalwareBazaar and MISP index by |
| source_url | URL the archive was downloaded from, where it is known |
| note | set on 34 entries that are not usable kits: empty (0 bytes), not-an-archive (no archi, mostly sa pages) and no-files (a valid zip containing no files) |
| url_pro | how source_url was obtained: recorded if it is the URL the download actually came from, inferred if it was reconstructed by the rule below, empty if unknown |

Nothing marked in note has been deleted, so existing references stay valid.

Reconstructed source URLs

Most of the archi any automated capture, so source_url was empty
for 899 of the entries. 833 of those were reconstructed from the per-month
urls.txt files and are marked inferred.

The rule: same month, the URL's host equals the host in the directory name
(tolerating a leading www.), and the URL's last path segment equals the
archive's filename, both percent-decoded. A row is only filled when exactly one
line in that month's urls.txt satisfies all three.

It was before being applied. Run against the entries that already had
a recorded URL, it reproduces that URL exactly in 190 of 190 cases, and on the
4 it cannot resolve it declines rather than guessing.

The 66 entries still empty are 42 where more than one line matched and 24 in a
month with no urls.txt. They were left empty on purpose. The e is
published, so the join can be re-derived, extended or disputed.

Look up a hash without cloning anything:

Metadata without cloning

The repository is around 3 GB and almost all of it is content rather than
history, so --depth 1 sa. What does help:

The per-month urls.txt files list the source URLs seen that month. They are not
an index: a line carries no reference to an archive, and matching one to the
other takes the reconstruction described above. Use index.csv instead, which
already holds the result.

Rights

The archives are third-party code, written by the phishing operators. No licence
is granted or implied over them and all rights remain with their respective
holders. What this repository offers is the collection, the layout and the
metadata in index.csv. Research and OSINT use only.

Remo

If you are the owner of data inside a kit, a brand being impersonated, or the
owner of a compromised host, see SECURITY.md. Report it pri
rather than opening a public issue, which would republish the exact path you are
asking to ha.

项目信息

开发语言：未知

Star 数：278

技术要点

该项目属于病毒分析方向的开源资源，代码结构清晰，适合安全研究人员学习参考。项目围绕Archive of phishing kits captured from suspicious domains detected by phishunt.io, with a sha256 manifest展开，提供了可运行的实现，对理解相关安全机制、工具原理与攻防技术具有较高的研究价值。

从技术实现角度看，此类项目的核心在于对底层原理的深入理解与工程化落地。开发者在研究过程中可以结合官方文档与社区讨论，梳理项目的初始化流程、核心模块职责与数据流转，逐步掌握其中的技术细节与设计思路。

使用与学习建议

1. 下载源码压缩包，解压到本地环境。

2. 根据项目类型准备对应的运行环境与依赖。

3. 按照 README 中的步骤完成配置与部署。

4. 在隔离的测试环境中运行验证，避免影响生产系统。

免责声明

本资源来源于开源社区，仅供安全研究、技术学习与合法授权范围内的攻防演练使用，请勿用于任何非法或侵害他人权益的用途。使用本项目产生的一切后果由使用者自行承担。