A flaw in how restforce constructs URLs may allow an attacker to inject additional parameters into Salesforce API requests.
This flaw is only exploitable in applications that pass user input directly to restforce's select, find, describe, update, upsert, and destroy methods.
Vulnerable code might look like:
client.select('SomeSalesForceObject', params[:some-id],
...)
In such an application, attackers could pass 0016000000MRatd/describe
as a request parameter, causing the server to make a request to a different endpoint than the server is designed to handle. Since the Salesforce REST API supports overriding HTTP methods via a request parameter, an attacker could also cause the client's select()
method to modify data, by passing 0016000000MRatd/?_HttpMethod=PATCH&other-query-params=...
.
If possible, applications should track salesforce IDs internally, rather than passing user-supplied IDs to salesforce. Such practice mitigates this vulnerability, and in general is desirable for ensuring strong access control.
{ "nvd_published_at": "2018-08-03T20:29:00Z", "cwe_ids": [ "CWE-172", "CWE-20" ], "severity": "CRITICAL", "github_reviewed": true, "github_reviewed_at": "2020-06-16T20:59:51Z" }