brom
v1.0.7
Published
brom is a configurable CLI for recording HTTP transactions and improving security practices, designed for use in local environments and CI tools. Get your headers in order _before_ deployment.
Downloads
5
Readme
brom
brom is a configurable CLI for recording HTTP transactions and improving security practices, designed for use in local environments and CI tools. Get your headers in order before deployment.
Table of Contents
Installation
npm install -g brom
Demo
This repository has a built-in brom config, so just clone, install, and run to see brom in action. You'll need Node 6.4.0 or later, and npm.
- Clone this repository locally
- In the root directory,
npm install
npm install -g brom
brom
- brom will open the GUI in your default browser automatically. Open a second tab and navigate to localhost:9999 to start making/recording requests.
Usage
brom <server.js> <port> [OPTIONS]
brom records HTTP transactions, generates configurable warnings, and provides you with detailed information on both requests and responses. By default, it opens a second browser tab to aid inspection with by-header breakdowns and a variety of filters.
If you'd rather go without the GUI and just get the data:
brom -g --write-dir=brom
writes all transactions from a given session to a timestamped file, assessments included.
brom recording sessions operate through two modes (which can run simultaneously):
Interactive mode (on by default)
brom proxies your server, recording HTTP transactions as they stream through, and injecting a script into HTML documents to record third-party AJAX calls.
Automated mode (Express.js only)
brom -a
brom traverses Express application structures to identify all registered routes and their supported methods, sending dumb requests to each one. Great for check-ins and (with a bit of configuration) build processes.
Rule syntax
While brom comes with smart defaults for assessing header practices, the security needs of your application are unique. brom provides a rule syntax to allow you to create your own rules. Here's a built-in rule:
{
id: 'no-x-powered-by', // linter name
header: 'x-powered-by', // used to organize warnings in GUI
when: (headers, type) => type === 'response', // when to run this rule
expect: headers => !headers['x-powered-by'], // pass/fail statement
fail: {
message: `The X-Powered-By header advertises unnecessary
details about your server, and should not be sent.`
}
}
The id
field is meant to uniquely identify rules, allowing you to tell
brom what rules to skip, and what rules should halt a build process on
failure. See Configuring brom.
when
Your when
callback helps brom decide on which transactions to run which
tests. brom passes it a headers object, with each header from an HTTP
message, and a type string to identify the message as a request or
response.
For instance, when Content-Security-Policy headers only matter on HTML responses:
{
// ...
when: (headers, type) => (
type === 'response' &&
headers['content-type'].match('text/html');
),
// ...
}
expect
The expect
callback represents your actual requirements: brom takes the
truthiness of the return value to indicate whether or not the transaction
failed this rule. It passes the same headers object as then when
callback.
Here's the expect callback for the same rule, to confirm that a CSP has been provided:
{
// ...
expect: headers => headers['feature-policy'],
// ...
}
Configuring brom
brom needs, at minimum, to know what local port to proxy to, and with
many settings it also needs a path to your server. If you'd rather just
type brom
, provide a brom.config.json
in your project's root
directory.
{
"server": "demo/server.js",
"port": 3000
}
All CLI flags can be preconfigured here as well. In a defaults
key
provide settings matching the longform of any flag. These settings will
always be overridden by flags and arguments passed directly to the CLI:
{
"server": "demo/server.js",
"port": 3000,
"defaults": {
"automated": true,
"https": true,
"preserve-caching": true,
"write-dir": "brom-history"
}
}
brom uses two lightweight servers to operate (a proxy for inspecting transactions, and a server for aggregating results from multiple sources). These operate on ports 9999 and 7913 respectively, but you can move them where you like:
{
"proxy-port": 3001,
"results-port": 3002
}
Some rules may not be relevant enough to your application to keep getting reminders. To turn these off in your config, add an array of rule ids to ignore:
{
"ignore-rules": [
"no-x-powered-by"
]
}
Conversely, to tell brom to exit with a nonzero status code on certain rules, add them to a "halt-on" key:
{
"halt-on": [
"no-x-powered-by",
"csp-no-*"
]
}
If you've configured your server to run HTTPS locally, brom needs to serve
HTTPS too. brom can piggyback off of the same key and certificate. Add
relative paths to the key and certificate, and fire up brom with the
--https
flag.
{
"https": {
"key": "../application/server.key",
"cert":"../application/server.crt"
}
}
API
When writing rules, some headers are tougher to parse, so brom provides a simple set of header parsers.
brom.contentSecurityPolicy(string)
Parses a CSP value and returns an object, with each directive name as a key and each directive value as an array of strings.
The value string must be the actual header value (everything after the
:
).
brom.csp(string)
Alias for brom.contentSecurityPolicy
.
brom.csp("default-src 'self'; img-src *; script-src userscripts.example.com");
// {
// 'default-src': ['\'self\''],
// 'img-src': ['*'],
// 'script-src': ['userscripts.example.com']
// }
brom.featurePolicy(string)
Parses a Feature Policy value and returns an object. Each key is a feature, and its value an array of whitelisted domains.
The value string must be the actual header value (everything after the
:
).
brom.fp(string)
Alias for brom.featurePolicy
.
brom.fp("vibrate 'none'; geolocation 'none'");
// {
// vibrate: ['\'none\''],
// geolocation: ['\'none\'']
// }
brom.cookie(string)
Parses a Cookie
value and returns an array of cookies. The cookie name and
value are separated into name
and value
keys.
The value string must be the actual header value (everything after the
:
).
brom.cookie('hello=world; foo=bar;');
// [
// { name: 'hello', value: 'world' },
// { name: 'foo', value: 'bar' }
// ]
brom.setCookie(string)
Parses a Set-Cookie
value and returns an object. The cookie name and
value are separated into name
and value
keys; the rest are separated
into key-value pairs (non-value directives like "Secure" are set to
true
).
The value string must be the actual header value (everything after the
:
).
brom.setCookie('a=b; Path=/; Max-Age=0; Secure; HttpOnly');
// [
// {
// name: 'a',
// value: 'b',
// Path: '/',
// 'Max-Age': '0',
// Secure: true,
// HttpOnly: true
// }
// ]
brom(object)
Returns a new object, with Content-Security-Policy
, Feature-Policy
,
Set-Cookie
, and Cookie
headers parsed. All other headers are left
unchanged under a single headers
key.
const parse = require('brom');
parse({
'Content-Security-Policy': "default-src 'self'; img-src *; media-src media1.com media2.com; script-src userscripts.example.com",
'Feature-Policy': "vibrate 'none'; geolocation 'none'",
'Cookie': 'hello=world; foo=bar;',
'Set-Cookie': 'a=b; Path=/; HttpOnly;',
'Content-Type': 'text/html; charset=utf-8',
'Content-Length': '7913'
});
// {
// contentSecurityPolicy: {
// 'default-src': ['\'self\''],
// 'img-src': ['*'],
// 'media-src': ['media1.com', 'media2.com'],
// 'script-src': ['userscripts.example.com']
// },
// cookies: [
// { name: 'hello', value: 'world' },
// { name: 'foo', value: 'bar' }
// ],
// featurePolicy: {
// vibrate: ['\'none\''],
// geolocation: ['\'none\'']
// },
// setCookie: [
// {
// name: 'a',
// value: 'b',
// Path: '/',
// 'Max-Age': '0',
// Secure: true,
// HttpOnly: true
// }
// ],
// headers: {
// 'Content-Type': 'text/html; charset=utf-8',
// 'Content-Length': '7913'
// }
// }
The headers object you pass in can contain any number of headers. Each
key should be a header name (everything up to the :
), and its
corresponding value the remainder of that header string.
The only exception to this rule is your Set-Cookie
header, which may be
either a string (representing a single cookie) or an array of strings
representing multiple cookies.
FAQ
Is brom for Express only?
Yes and no: brom hasn't been tested extensively on non-Node backends, but the interactive-mode flow has no interaction with or awareness of your server, other than the port to proxy to.
However, brom's automated mode is designed explicitly for Express servers
(off by default), and you'll need to turn off autostart (add an -m
flag)
to tell brom not to look for a server file. Start up your server as its
own process, and run brom <port> -m
.
Does brom support HTTPS?
As long as you've set up local HTTPS before, brom can piggyback off of the same certificate. See the Configuration section for more details on configuring HTTPS.