agentleFS
ajtazerajtazer/heckit/skills/cve-hunting/entity-expansion/SKILL.md

entity-expansion

Skill2 starsChanged 2 months agoMITView on GitHub

Detect XML/SVG/YAML entity expansion (Billion Laughs) vulnerabilities in parsers that allow unbounded entity definitions.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Document processors (DOCX, XLSX, RSS, Atom, SOAP)

    line 23

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Entity Expansion (Billion Laughs) Detection

When to Use

Audit any package that parses XML, SVG, HTML with entity support, or YAML with alias/anchor support. This includes:

  • XML/SVG parsing libraries
  • Document processors (DOCX, XLSX, RSS, Atom, SOAP)
  • YAML parsers with alias expansion
  • Configuration file parsers

~90% CVE acceptance rate when confirmed.

Key Insight

Many parsers have NO default entity expansion limit. A 1KB XML payload with recursive entity definitions can expand to 1GB+ in memory, crashing the process with an OOM kill (uncatchable — process dies).

Entity Expansion Types

1. Billion Laughs (Internal Entity Recursion)

<?xml version="1.0"?>
<!DOCTYPE lolz [
  <!ENTITY lol "lol">
  <!ENTITY lol2 "&lol;&lol;&lol;&lol;&lol;&lol;&lol;&lol;&lol;&lol;">
  <!ENTITY lol3 "&lol2;&lol2;&lol2;&lol2;&lol2;&lol2;&lol2;&lol2;&lol2;&lol2;">
  ...
  <!ENTITY lol9 "&lol8;&lol8;&lol8;&lol8;&lol8;&lol8;&lol8;&lol8;&lol8;&lol8;">
]>
<root>&lol9;</root>

Each level multiplies by 10. Level 9 = 10^9 = 1 billion "lol" strings.

2. Quadratic Blowup (Single Entity Repeated)

<!DOCTYPE foo [
  <!ENTITY a "AAAAAAAAAA..."> <!-- 50KB entity -->
]>
<root>&a;&a;&a;&a;&a;...&a;</root> <!-- 50000 references -->

Less dramatic but still effective — 50KB entity × 50000 refs = 2.5GB.

3. YAML Alias Expansion

a: &anchor
  x: *anchor
  y: *anchor

Recursive alias references can cause exponential expansion in some YAML parsers.

Process

Step 1: Find XML/YAML Parsing

# JavaScript
grep -rn "xml2js\|fast-xml-parser\|xmldom\|sax\|DOMParser\|cheerio" .
grep -rn "\.parseString\|\.parse(" . --include="*.js" --include="*.ts"
grep -rn "yaml\.load\|yaml\.parse\|YAML\.parse" .

# Python
grep -rn "xml\.etree\|lxml\|minidom\|xml\.sax\|defusedxml" .
grep -rn "yaml\.load\|yaml\.safe_load\|yaml\.unsafe_load" .

# Go
grep -rn "xml\.Decoder\|xml\.Unmarshal\|encoding/xml" .
grep -rn "yaml\.Unmarshal\|gopkg.in/yaml" .

# Ruby
grep -rn "Nokogiri\|REXML\|Ox\.\|LibXML" .

# PHP
grep -rn "simplexml\|DOMDocument\|XMLReader\|xml_parse" .

Step 2: Check Entity/DTD Configuration

For each parser found, check:

  1. Does it process DTD declarations by default?
  2. Is there a maxExpansion or maxEntitySize option?
  3. Is DTD processing explicitly disabled?
  4. Does it support custom entity resolution?

Step 3: Check for Expansion Limits

grep -rn "maxExpansion\|maxEntitySize\|entityExpansion\|ENTITY_EXPANSION" .
grep -rn "disableDTD\|forbidDTD\|dtd.*false\|FEATURE_SECURE_PROCESSING" .
grep -rn "noent\|resolve_entities\|processEntities" .

Step 4: Check YAML Alias Limits

grep -rn "maxAliasCount\|maxAliases\|aliasLimit\|MAX_ALIAS" .
grep -rn "anchorLimit\|maxAnchors" .

Step 5: Verify Exploitability

  1. Does the parser accept untrusted input? (user uploads, API requests, webhook payloads)
  2. Is there a file size limit that would prevent the payload from being processed?
  3. Is the process memory-limited (cgroups, ulimit)?
  4. Does the parser use streaming that could limit memory usage?

Known Parser Default Safety

ParserLanguageDTD/Entity DefaultSafe?
fast-xml-parserJSEntities processed, no limitUNSAFE
xml2jsJSEntities processed, no limitUNSAFE
xmldomJSEntities processed, no limitUNSAFE
saxJSNo entity expansionSAFE
cheerio (htmlparser2)JSNo DTD supportSAFE
xml.etree.ElementTreePythonEntities processed, no limitUNSAFE
lxmlPythonDTD disabled by defaultSAFE (default)
defusedxmlPythonAll dangerous features disabledSAFE
xml.saxPythonEntities processedUNSAFE
PyYAML yaml.loadPythonAliases processed, no limitUNSAFE
PyYAML yaml.safe_loadPythonAliases processed, no limitUNSAFE (aliases)
encoding/xmlGoNo entity supportSAFE
go-yaml v3GoAliases processed, limitedCHECK VERSION
NokogiriRubyDTD disabled by defaultSAFE (default)
REXMLRubyEntities processedUNSAFE
simplexml_load_stringPHPEntities processed by defaultUNSAFE
DOMDocumentPHPEntities processed by defaultUNSAFE

CVSS Guidance

  • Unauthenticated OOM crash (process dies): HIGH 7.5 (AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H)
  • Authenticated OOM crash: MEDIUM 6.5 (PR:L)
  • CPU exhaustion (recoverable): MEDIUM 5.3

References

  • Sinks — XML/YAML parser safety status by language
  • False Positive Indicators — When this isn't exploitable
  • PoC Skeleton — Billion Laughs payload templates

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.