Node.js v22.0.0-v8-canary20231031532aa30be7 documentation

Module-based permissions#

Policies#

Node.js contains experimental support for creating policies on loading code.

Policies are a security feature intended to ensure the integrity of the loaded code.

While it does not function as a provenance mechanism to trace the origin of code, it serves as a robust defense against the execution of malicious code. Unlike runtime-based models that may restrict capabilities once the code is loaded, Node.js policies focus on preventing malicious code from ever being fully loaded into the application in the first place.

The use of policies assumes safe practices for the policy files such as ensuring that policy files cannot be overwritten by the Node.js application by using file permissions.

A best practice would be to ensure that the policy manifest is read-only for the running Node.js application and that the file cannot be changed by the running Node.js application in any way. A typical setup would be to create the policy file as a different user id than the one running Node.js and granting read permissions to the user id running Node.js.

Enabling#

The --experimental-policy flag can be used to enable features for policies when loading modules.

Once this has been set, all modules must conform to a policy manifest file passed to the flag:

node --experimental-policy=policy.json app.js copy

The policy manifest will be used to enforce constraints on code loaded by Node.js.

To mitigate tampering with policy files on disk, an integrity for the policy file itself may be provided via --policy-integrity. This allows running node and asserting the policy file contents even if the file is changed on disk.

node --experimental-policy=policy.json --policy-integrity="sha384-SggXRQHwCG8g+DktYYzxkXRIkTiEYWBHqev0xnpCxYlqMBufKZHAHQM3/boDaI/0" app.js copy

Features#

Error behavior#

When a policy check fails, Node.js by default will throw an error. It is possible to change the error behavior to one of a few possibilities by defining an "onerror" field in a policy manifest. The following values are available to change the behavior:

• "exit": will exit the process immediately. No cleanup code will be allowed to run.
• "log": will log the error at the site of the failure.
• "throw": will throw a JS error at the site of the failure. This is the default.

{
  "onerror": "log",
  "resources": {
    "./app/checked.js": {
      "integrity": "sha384-SggXRQHwCG8g+DktYYzxkXRIkTiEYWBHqev0xnpCxYlqMBufKZHAHQM3/boDaI/0"
    }
  }
} copy

Integrity checks#

Policy files must use integrity checks with Subresource Integrity strings compatible with the browser integrity attribute associated with absolute URLs.

When using require() or import all resources involved in loading are checked for integrity if a policy manifest has been specified. If a resource does not match the integrity listed in the manifest, an error will be thrown.

An example policy file that would allow loading a file checked.js:

{
  "resources": {
    "./app/checked.js": {
      "integrity": "sha384-SggXRQHwCG8g+DktYYzxkXRIkTiEYWBHqev0xnpCxYlqMBufKZHAHQM3/boDaI/0"
    }
  }
} copy

Each resource listed in the policy manifest can be of one the following formats to determine its location:

1. A relative-URL string to a resource from the manifest such as ./resource.js, ../resource.js, or /resource.js.
2. A complete URL string to a resource such as file:///resource.js.

When loading resources the entire URL must match including search parameters and hash fragment. ./a.js?b will not be used when attempting to load ./a.js and vice versa.

To generate integrity strings, a script such as node -e 'process.stdout.write("sha256-");process.stdin.pipe(crypto.createHash("sha256").setEncoding("base64")).pipe(process.stdout)' < FILE can be used.

Integrity can be specified as the boolean value true to accept any body for the resource which can be useful for local development. It is not recommended in production since it would allow unexpected alteration of resources to be considered valid.

Dependency redirection#

An application may need to ship patched versions of modules or to prevent modules from allowing all modules access to all other modules. Redirection can be used by intercepting attempts to load the modules wishing to be replaced.

{
  "resources": {
    "./app/checked.js": {
      "dependencies": {
        "fs": true,
        "os": "./app/node_modules/alt-os",
        "http": { "import": true }
      }
    }
  }
} copy

The dependencies are keyed by the requested specifier string and have values of either true, null, a string pointing to a module to be resolved, or a conditions object.

The specifier string does not perform any searching and must match exactly what is provided to the require() or import except for a canonicalization step. Therefore, multiple specifiers may be needed in the policy if it uses multiple different strings to point to the same module (such as excluding the extension).

Specifier strings are canonicalized but not resolved prior to be used for matching in order to have some compatibility with import maps, for example if a resource file:///C:/app/server.js was given the following redirection from a policy located at file:///C:/app/policy.json:

{
  "resources": {
    "file:///C:/app/utils.js": {
      "dependencies": {
        "./utils.js": "./utils-v2.js"
      }
    }
  }
} copy

Any specifier used to load file:///C:/app/utils.js would then be intercepted and redirected to file:///C:/app/utils-v2.js instead regardless of using an absolute or relative specifier. However, if a specifier that is not an absolute or relative URL string is used, it would not be intercepted. So, if an import such as import('#utils') was used, it would not be intercepted.

If the value of the redirection is true, a "dependencies" field at the top of the policy file will be used. If that field at the top of the policy file is true the default node searching algorithms are used to find the module.

If the value of the redirection is a string, it is resolved relative to the manifest and then immediately used without searching.

Any specifier string for which resolution is attempted and that is not listed in the dependencies results in an error according to the policy.

A boolean value of true for the dependencies map can be specified to allow a module to load any specifier without redirection. This can be useful for local development and may have some valid usage in production, but should be used only with care after auditing a module to ensure its behavior is valid.

Similar to "exports" in package.json, dependencies can also be specified to be objects containing conditions which branch how dependencies are loaded. In the preceding example, "http" is allowed when the "import" condition is part of loading it.

A value of null for the resolved value causes the resolution to fail. This can be used to ensure some kinds of dynamic access are explicitly prevented.

Unknown values for the resolved module location cause failures but are not guaranteed to be forward compatible.

All the guarantees for policy redirection are specified in the Guarantees section.

Example: Patched dependency#

Redirected dependencies can provide attenuated or modified functionality as fits the application. For example, log data about timing of function durations by wrapping the original:

const original = require('fn');
module.exports = function fn(...args) {
  console.time();
  try {
    return new.target ?
      Reflect.construct(original, args) :
      Reflect.apply(original, this, args);
  } finally {
    console.timeEnd();
  }
}; copy

Scopes#

Use the "scopes" field of a manifest to set configuration for many resources at once. The "scopes" field works by matching resources by their segments. If a scope or resource includes "cascade": true, unknown specifiers will be searched for in their containing scope. The containing scope for cascading is found by recursively reducing the resource URL by removing segments for special schemes, keeping trailing "/" suffixes, and removing the query and hash fragment. This leads to the eventual reduction of the URL to its origin. If the URL is non-special the scope will be located by the URL's origin. If no scope is found for the origin or in the case of opaque origins, a protocol string can be used as a scope. If no scope is found for the URL's protocol, a final empty string "" scope will be used.

Note, blob: URLs adopt their origin from the path they contain, and so a scope of "blob:https://nodejs.org" will have no effect since no URL can have an origin of blob:https://nodejs.org; URLs starting with blob:https://nodejs.org/ will use https://nodejs.org for its origin and thus https: for its protocol scope. For opaque origin blob: URLs they will have blob: for their protocol scope since they do not adopt origins.

Example#

{
  "scopes": {
    "file:///C:/app/": {},
    "file:": {},
    "": {}
  }
} copy

Given a file located at file:///C:/app/bin/main.js, the following scopes would be checked in order:

1. "file:///C:/app/bin/"

This determines the policy for all file based resources within "file:///C:/app/bin/". This is not in the "scopes" field of the policy and would be skipped. Adding this scope to the policy would cause it to be used prior to the "file:///C:/app/" scope.

2. "file:///C:/app/"

This determines the policy for all file based resources within "file:///C:/app/". This is in the "scopes" field of the policy and it would determine the policy for the resource at file:///C:/app/bin/main.js. If the scope has "cascade": true, any unsatisfied queries about the resource would delegate to the next relevant scope for file:///C:/app/bin/main.js, "file:".

3. "file:///C:/"

This determines the policy for all file based resources within "file:///C:/". This is not in the "scopes" field of the policy and would be skipped. It would not be used for file:///C:/app/bin/main.js unless "file:///" is set to cascade or is not in the "scopes" of the policy.

4. "file:///"

This determines the policy for all file based resources on the localhost. This is not in the "scopes" field of the policy and would be skipped. It would not be used for file:///C:/app/bin/main.js unless "file:///" is set to cascade or is not in the "scopes" of the policy.

5. "file:"

This determines the policy for all file based resources. It would not be used for file:///C:/app/bin/main.js unless "file:///" is set to cascade or is not in the "scopes" of the policy.

6. ""

This determines the policy for all resources. It would not be used for file:///C:/app/bin/main.js unless "file:" is set to cascade.

Integrity using scopes#

Setting an integrity to true on a scope will set the integrity for any resource not found in the manifest to true.

Setting an integrity to null on a scope will set the integrity for any resource not found in the manifest to fail matching.

Not including an integrity is the same as setting the integrity to null.

"cascade" for integrity checks will be ignored if "integrity" is explicitly set.

The following example allows loading any file:

{
  "scopes": {
    "file:": {
      "integrity": true
    }
  }
} copy

Dependency redirection using scopes#

The following example, would allow access to fs for all resources within ./app/:

{
  "resources": {
    "./app/checked.js": {
      "cascade": true,
      "integrity": true
    }
  },
  "scopes": {
    "./app/": {
      "dependencies": {
        "fs": true
      }
    }
  }
} copy

The following example, would allow access to fs for all data: resources:

{
  "resources": {
    "data:text/javascript,import('node:fs');": {
      "cascade": true,
      "integrity": true
    }
  },
  "scopes": {
    "data:": {
      "dependencies": {
        "fs": true
      }
    }
  }
} copy

Example: import maps emulation#

Given an import map:

{
  "imports": {
    "react": "./app/node_modules/react/index.js"
  },
  "scopes": {
    "./ssr/": {
      "react": "./app/node_modules/server-side-react/index.js"
    }
  }
} copy

{
  "dependencies": true,
  "scopes": {
    "": {
      "cascade": true,
      "dependencies": {
        "react": "./app/node_modules/react/index.js"
      }
    },
    "./ssr/": {
      "cascade": true,
      "dependencies": {
        "react": "./app/node_modules/server-side-react/index.js"
      }
    }
  }
} copy

Import maps assume you can get any resource by default. This means "dependencies" at the top level of the policy should be set to true. Policies require this to be opt-in since it enables all resources of the application cross linkage which doesn't make sense for many scenarios. They also assume any given scope has access to any scope above its allowed dependencies; all scopes emulating import maps must set "cascade": true.

Import maps only have a single top level scope for their "imports". So for emulating "imports" use the "" scope. For emulating "scopes" use the "scopes" in a similar manner to how "scopes" works in import maps.

Caveats: Policies do not use string matching for various finding of scope. They do URL traversals. This means things like blob: and data: URLs might not be entirely interoperable between the two systems. For example import maps can partially match a data: or blob: URL by partitioning the URL on a / character, policies intentionally cannot. For blob: URLs import map scopes do not adopt the origin of the blob: URL.

Additionally, import maps only work on import so it may be desirable to add a "import" condition to all dependency mappings.

Guarantees#

• The policies guarantee the file integrity when a module is loaded using require(), import() or new Module().
• Redirection does not prevent access to APIs through means such as direct access to require.cache which allow access to loaded modules. Policy redirection only affects specifiers to require() and import.
• The approval of the module integrity in policies threat model implies they are allowed to muck with and even circumvent security features once loaded so environmental/runtime hardening is expected.

Process-based permissions#

Permission Model#

The Node.js Permission Model is a mechanism for restricting access to specific resources during execution. The API exists behind a flag --experimental-permission which when enabled, will restrict access to all available permissions.

The available permissions are documented by the --experimental-permission flag.

When starting Node.js with --experimental-permission, the ability to access the file system through the fs module, spawn processes, use node:worker_threads and enable the runtime inspector will be restricted.

$ node --experimental-permission index.js
node:internal/modules/cjs/loader:171
  const result = internalModuleStat(filename);
                 ^

Error: Access to this API has been restricted
    at stat (node:internal/modules/cjs/loader:171:18)
    at Module._findPath (node:internal/modules/cjs/loader:627:16)
    at resolveMainPath (node:internal/modules/run_main:19:25)
    at Function.executeUserEntryPoint [as runMain] (node:internal/modules/run_main:76:24)
    at node:internal/main/run_main_module:23:47 {
  code: 'ERR_ACCESS_DENIED',
  permission: 'FileSystemRead'
} copy

Allowing access to spawning a process and creating worker threads can be done using the --allow-child-process and --allow-worker respectively.

Runtime API#

When enabling the Permission Model through the --experimental-permission flag a new property permission is added to the process object. This property contains one function:

permission.has(scope[, reference])#

API call to check permissions at runtime (permission.has())

process.permission.has('fs.write'); // true
process.permission.has('fs.write', '/home/rafaelgss/protected-folder'); // true

process.permission.has('fs.read'); // true
process.permission.has('fs.read', '/home/rafaelgss/protected-folder'); // false copy

File System Permissions#

To allow access to the file system, use the --allow-fs-read and --allow-fs-write flags:

$ node --experimental-permission --allow-fs-read=* --allow-fs-write=* index.js
Hello world!
(node:19836) ExperimentalWarning: Permission is an experimental feature
(Use `node --trace-warnings ...` to show where the warning was created) copy

The valid arguments for both flags are:

• * - To allow all FileSystemRead or FileSystemWrite operations, respectively.
• Paths delimited by comma (,) to allow only matching FileSystemRead or FileSystemWrite operations, respectively.

Example:

• --allow-fs-read=* - It will allow all FileSystemRead operations.
• --allow-fs-write=* - It will allow all FileSystemWrite operations.
• --allow-fs-write=/tmp/ - It will allow FileSystemWrite access to the /tmp/ folder.
• --allow-fs-read=/tmp/ --allow-fs-read=/home/.gitignore - It allows FileSystemRead access to the /tmp/ folder and the /home/.gitignore path.

Wildcards are supported too:

• --allow-fs-read=/home/test* will allow read access to everything that matches the wildcard. e.g: /home/test/file1 or /home/test2

Limitations and known issues#

There are constraints you need to know before using this system:

• When the permission model is enabled, Node.js may resolve some paths differently than when it is disabled.
• Native modules are restricted by default when using the Permission Model.
• OpenSSL engines currently cannot be requested at runtime when the Permission Model is enabled, affecting the built-in crypto, https, and tls modules.
• Relative paths are not supported through the CLI (--allow-fs-*).
• The model does not inherit to a child node process.
• The model does not inherit to a worker thread.
• Symbolic links will be followed even to locations outside of the set of paths that access has been granted to. Relative symbolic links may allow access to arbitrary files and directories. When starting applications with the permission model enabled, you must ensure that no paths to which access has been granted contain relative symbolic links.
• When creating symlinks the target (first argument) should have read and write access.
• Permission changes are not retroactively applied to existing resources.
• The Permission Model is initialized after the Node.js environment is set up. However, certain flags such as --env-file or --openssl-config are designed to read files before environment initialization. As a result, such flags are not subject to the rules of the Permission Model.