This package is the core of CASL. It includes logic responsible for checking and defining permissions.
npm install @casl/ability
# or
pnpm install @casl/ability
# or
yarn add @casl/abilityThis README file contains only basic information about the package. If you need an in depth introduction, please visit CASL's documentation.
Note: the best way to get started is to read the Guide in the official documentation. In this README file, you will find just basic information.
Note: all the examples below are written in ES6 using ES modules but CASL also has a sophisticated support for TypeScript, read CASL TypeScript support for details.
Lets define Ability for a blog website where visitors:
- can read blog posts
- can manage (i.e., do anything) own posts
- cannot delete a post if it was created more than a day ago
import { AbilityBuilder, createMongoAbility } from '@casl/ability';
import { User } from '../models'; // application specific interfaces
/**
* @param user contains details about logged in user: its id, name, email, etc
*/
function defineAbilitiesFor(user) {
const { can, cannot, build } = new AbilityBuilder(createMongoAbility);
// can read blog posts
can('read', 'BlogPost');
// can manage (i.e., do anything) own posts
can('manage', 'BlogPost', { author: user.id });
// cannot delete a post if it was created more than a day ago
cannot('delete', 'BlogPost', {
createdAt: { $lt: Date.now() - 24 * 60 * 60 * 1000 }
});
return build();
});Do you see how easily business requirements were translated into CASL's rules?
And yes, Ability class allow you to use some MongoDB operators to define conditions. Don't worry if you don't know MongoDB, it's not required and explained in details in Defining Abilities
Later on you can check abilities by using can and cannot methods of Ability instance.
import { BlogPost, ForbiddenError } from '../models';
const user = getLoggedInUser(); // app specific function
const ability = defineAbilitiesFor(user)
// true if ability allows to read at least one Post
ability.can('read', 'BlogPost');
// true if there is no ability to read this particular blog post
const post = new BlogPost({ title: 'What is CASL?' });
ability.cannot('read', post);
// you can even throw an error if there is a missed ability
ForbiddenError.from(ability).throwUnlessCan('read', post);Of course, you are not restricted to use only class instances in order to check permissions on objects. See Introduction for the detailed explanation.
CASL has a complementary package [@casl/mongoose] which provides easy integration with MongoDB and [mongoose].
import { accessibleRecordsPlugin } from '@casl/mongoose';
import mongoose from 'mongoose';
mongoose.plugin(accessibleRecordsPlugin);
const user = getUserLoggedInUser(); // app specific function
const ability = defineAbilitiesFor(user);
const BlogPost = mongoose.model('BlogPost', mongoose.Schema({
title: String,
author: mongoose.Types.ObjectId,
content: String,
createdAt: Date,
hidden: { type: Boolean, default: false }
}))
// returns mongoose Query, so you can chain it with other conditions
const posts = await Post.accessibleBy(ability).where({ hidden: false });
// you can also call it on existing query to enforce permissions
const hiddenPosts = await Post.find({ hidden: true }).accessibleBy(ability);
// you can even pass the action as a 2nd parameter. By default action is "read"
const updatablePosts = await Post.accessibleBy(ability, 'update');See Database integration for details.
CASL is incrementally adoptable, that means you can start your project with simple claim (or action) based authorization and evolve it later, when your app functionality evolves.
CASL is composable, that means you can implement alternative conditions matching (e.g., based on joi, ajv or pure functions) and field matching (e.g., to support alternative syntax in fields like addresses.*.street or addresses[0].street) logic.
See Advanced usage for details.
CASL checks are quite fast, thanks to underlying rule index structure. The estimated complexity of different operations can be found below:
| Operation | Complexity | Notes |
|---|---|---|
Ability creation time |
O(n) | n - amount of rules |
Check by action and subject type (e.g., ability.can('read', 'Todo')) |
O(1) | |
Check by action and subject object (e.g., ability.can('read', todo)) |
O(m + k) + O(p) | m - amount of rules for the same pair of action and subject; k - amount of operators in conditions; O(p) - complexity of used operators (e.g., $in implementation is more complex than $lt) |
Want to file a bug, contribute some code, or improve documentation? Excellent! Read up on guidelines for contributing