Joint-assessment Outline

The joint-assessment is built on top of the self-assessment.md to collaboratively assess the current security state of a project.

The burden is primarily on the proposing project to demonstrate it is secure in a manner that is understandable to the broader community. The reviewers will help to assess and probe the design.

The proposing project must provide a written document that describes the project and its security. In the case of OpenFGA, there is structured information present in the Security-Insights page. The project self assessment has been completed.

Projects are encouraged to cross link additional supporting documents or details from their repo into the self-assessment.

Joint-assessment of OpenFGA

Table of Contents

• Metadata
  • Security links
• Overview
  • Background
  • Goals
  • Non-goals
• Joint-assessment use
• Intended use
• Project design
  • Functions and features
    • Security functions and features
• Configuration and set-up
• Project compliance
• Security analysis
• Secure development practices
• Security issue resolution
  • Closed security issues and vulnerabilities
• Hands-on assessment
• Roadmap
• Appendix

Metadata

A table at the top for quick reference information, later used for indexing.

Security links

These are link to existing security documentation for the project.

Overview

The overview sections are pulled from the self-assessment and updated.

Implementing access control is a very common requirement when developing applications, where different subjects can perform different actions on different resources.

OpenFGA is a high performance and flexible authorization/permission engine that can be used to implement fine grained access control in any application component.

Developers can use OpenFGA to easily craft authorization and permission methods based on the policies they require that are specific to their own projects. They can further use the API’s provided by the project to confirm users have the permissions required to access a given resource.

Background

OpenFGA is an authorization/permission engine that incorporates Relationship-Based Access Control (ReBAC) and Attribute Based Access Control (ABAC) concepts with a domain-specific language that enables crafting authorizations solutions that can grow and evolve to any use case.

It’s inspired on the idea described in the Google Zanzibar paper .

Fine-Grained Authorization refers to individual users having access to specific objects and resources within a system. Google Drive is an example of this, as owners of resources can grant different users to have different levels of access to their resources.

OpenFGA makes helps developers make authorization decisions by combining two concepts:

• An Authorization Model, where developers define their authorization policies

• A set of relationship tuples that instantiate the model and OpenFGA uses to answer access control queries.

An authorization model looks like:

model
  schema 1.1

type user
type group
  relations
    define member: [user]
type folder
  relations
    define owner: [user]
    define parent: [folder]
    define viewer: [user, group#member] or owner or viewer from parent

type document
  relations
    define parent: [folder]
    define owner: [user]
    define viewer: [user, group#member] or owner or viewer from parent

Relationship tuples look like:

With this information, OpenFGA can be queried in different ways:

• Using the /check endpoint to ask questions like “Is user:alice a viewer for document:readme?”. With the data provided above, OpenFGA will return {allowed : "true"}, as Alice is a member of the engineering team, which has viewer access on the ‘readme’ document’s parent folder.

• Using the /list-objects endpoint to ask questions like “What are all the documents for which user:alice is a viewer. With the data provided above, OpenFGA will return {object_ids { "document:readme" }

Goal

• Simplify and standardize authorization processes, making them more consistent across various applications and systems.

• Establish patterns and standards for externalized authorization.

• Create architectural patterns, terminologies, and protocols that enable interoperability among different authorization systems.

• Deliver an authorization service for any application component.

• Enable centralized authorization decisions and permits diverse teams to implement authorization using a shared framework across various application components.

Non-goals

• Tools for management of groups/roles/permissions not inherently provided to the end-users.

• Does not intend to serve as a comprehensive data repository for non-authorization related data.

• Does not aim to provide a complete authentication and Access Control Solution.

Joint-assessment use

The joint-assessment is initially created by the project team and then collaboratively developed with the security reviewers as part of the project’s TAG-Security Security Assessment (TSSA) Process. Information about the TAG-Security Review can be found in the CNCF TAG-Security Review Process Guide .

This document does not intend to provide a security audit of OpenFGA and is not intended to be used in lieu of a security audit. This document provides users of the project with a security focused understanding of OpenFGA and when taken with the self-assessment provide the community with the TAG-Security Review of the project. Both of these documents may be used and referenced as part of a security audit.

OpenFGA is a project that provides a security service and as such, any defect in the project may be a security issue. This document does not look to enumerate all the possible quality issues (e.g. undetected circular references in model definitions) that could lead to security issues for users of OpenFGA.

Taken together, this document and the self-assessment serve as a cornerstone for if and when OpenFGA seeks graduation and is preparing for a security audit.

Intended Use

• Target Users and Use Cases. The key users of this project are users who define authorization models, application developers that integrate the API into their application for externalizing authorization and operators.

  OpenFGA can be used in any environment and has helm charts defined for install on a Kubernetes platform.

  1. OpenFGA is used by applications to externalize authorization decisions
  2. The project implements the Google Zanzibar paper paper for effective, performant authorization
  3. Administrators can program authorization models into the system for use by application teams

• Operation. OpenFGA supports both MySQL and Postgres as its datastore.

Project Design

• Design. OpenFGA provides rich documentation around its core concepts and usage of the project. Some project decisions are documented under the rfcs repository , however, this does not have a comprehensive list of all project decisions.

Functions and features

The list below describes the functionality provided by OpenFGA. This assessment did not segregate these into critical and other levels but focused on the server component as a priority.

actions:
  system.users:
    - Request access to [system.resources] through [openfga.clients|system.applications]

  system.developers:
    - Integrate [openfga.sdks] in [openfga.clients|system.applications]
    - Validate and verify semantically [openfga.authz_models]

  system.operators:
    - Migrate [openfga.datastore]
    - Deploy [openfga.server]

  system.external.idp:
    - Provide [jwks_uri] through oidc /.well-known/openid-configuration
    - Sign [token] with [rs256] algorithm

  openfga.language:
    - Provide a domain specific language to describe authorization policies
    - Describe the authorization model with [types], [relations] and [conditions]

  openfga.datastore:
    - Store authorization models [openfga.authz_models]
    - Store authorization data [openfga.relationships.tuples]
    - Support for [MySQL, Postgres] database

  openfga.clients|system.applications:
    - Authenticate against [openfga.server] with [openfga.psk] secret or through [external.idp]
    - Execute authorization checks with [openfga.relationships.queries]
    - Manage the authorization model [openfga.authz_models]

  openfga.server:
    - Write authorization model [openfga.authz_models] to [openfga.datastore]
    - Write authorization data [openfga.relationships.tuples] to [openfga.datastore]
    - Provide [grpc|http] messaging protocol
    - Authenticate trusted [openfga.clients] with 3 options [none|psk|oidc]
    - Validate and verify [payload]
    - Evaluate access control decisions [openfga.relationships.queries]

  openfga.server.api:
    stores:
      - list
      - create
      - get
      - delete
      - assertions.read
      - assertions.upsert
    authz-models:
      - list
      - create
      - get
    relationships.tuples:
      - read
      - write
      - delete
      - list.changes
    relationships.queries:
      - check
      - expand
      - list-objects
      - streamed-list-objects
      - list-users

Security functions and features

OpenFGA, being an Open-Source project, allows for a more robust security implementation by following the Principle of Open Design.

OpenFGA models authorization systems by providing the security features such as Role-based Access Control and Attribute-based Access Control.

OpenFGA boasts exceptional speed in processing secure authorization check call. This swift authorization mechanism not only enhances efficiency but also reinforces the security posture, assuring robust protection for applications and platforms for diverse scales.

OpenFGA provides a wide variety of SDK’s, as well as easy integration for new SDK’s. This reduces the chance of critical vulnerabilities due to compatibility issues.

Configuration and Set-Up

• Default. The default configuration of the project is described under the getting started section of the documentation.

• Secure. The recommended production configuration of the project is described under a separate part of the documentation.

Project Compliance

There are no specific security standards or sub-sections the project is documented as meeting (e.g. NIST 800-53, HiTrust, etc.).

Existing Audits

A self-assessment has been performed but there are no project audits currently for OpenFGA.

Security Analysis

Attacker Motivations

OpenFGA is an authorization/ permission engine that secures sensitive information including users of the systems referenced by the permission model as well as permissions and access levels for each. Attackers may have multiple motivations including:

• Exfiltrating data from the service
• Tampering with data to assign or elevate permissions
• Studying the data to understand the technology landscape
• Denial of service by rendering OpenFGA unable to respond to auth requests

Predisposing Conditions

There are multiple potential configurations of the project that could be exploited, this includes permissive settings for API tokens and other secrets, running the server in elevated/ privileged mode, exposing vulnerable endpoints accessible to an attacker and exfiltration of secret tokens from interfacing applications and CLI.

In addition vulnerabilities may be discovered in the server and other components that can be exploited by an attacker.

Expected Attacker Capabilities

While the attacker is not expected to possess the capability to break well-known encryption standards such as SHA256, they will have sufficient capabilities and motivation to use well-known tools and techniques for their work. Attackers are not just assumed to be external to the company but may also be persistent threats within the company network that are looking to gain a foothold for further exploits. The latter scenarios assume that the attacker has breached several layers of defense and has direct access to OpenFGA components and endpoints to further their position within the company perimeter.

Attack Risks and Effects

While not storing PII (as per project recommendations), OpenFGA does have sensitive information pertaining to the technology landscape and permissions within the organization. This data could be used by attackers to better understand the landscape and also perform more destructive actions such as escalating their permissions and locking out users from system access. In a microservices environment, data tampering may lead to subtle issues within the stack that are difficult to debug and may degrade customer experience and tie up technology teams in troubleshooting.

Security Degradation

A complete compromise of the OpenFGA server in production would lead to ripple effects throughout the organizations. Attackers could assign themselves arbitraty permissions to sensitive systems and also lock out legitimate users. This could lead to effects from a denial of service and degradation of customer experience to exfiltration of critical data from sensitive systems. OpenFGA deals directly with authorization and is a critical part of any organization, as such, compromising this system could have catastrophic consequences to the organization.

Compensating Mechanisms

Compensating mechanisms are covered in detail in the Threat Model section. These cover steps such as architecture changes for a more fine-grained permissions system, hardening the default deployment instructions, changes to user documentation and changes in functionality to address common exploits.

Threat Model

Threat modelling was done using MITRE Att&ck. Findings are listed below along with the Att&ck technique associated with the finding. The findings are categorized into logical sections. The values for the Impact and Likelihood ratings can be High, Med or Low.

Opportunities for improvement identified include:

• Implement FGA for server API
• Relook at user-defined API tokens as an authentication mechanism for API
• Make all installation scripts “secure by default”
• Validate best practices such as using strong API token and avoiding PII
• Look at SPIFFE/ Spire integration as an option for OpenFGA server

Further opportunities for improvement are listed under the Secure Development section.

Methods of authentication for server API

Access to OpenFGA API is via oauth or pre-determined API tokens. Pre-shared tokens present several weak points that result in the findings below.

The project recommends oauth for secure authentication to the API. The recommendation is to mark shared API tokens as a relatively insecure method of authentication and that his be avoided in a production environment. This recommendation can be updated both in the documentation as well as a WARNING can be emmitted in the logs when authentication via shared API tokens is enabled.

The recommendation to enable Fine Grained Authorization for the API is being implemented under a project issue . This risk is currently being mitigated by OpenFGA users by proxying requests and performing authorization on the proxy.

Setting secure defaults for install

This section has findings related to default install options that can be made more secure. The artifacts analysed in this section include the helm chart for installation and default configuration options.

Other findings

This section has other findings that could not be classified in earlier parts. It includes exploits such as server DDOS and potential leakage of information about application landscape.

Security Hygiene and Secure Development Practices

This section addresses questions related to project-level security decisions.

Opportunities for improvement include:

• Multi-organizational ownership of security reporting
• A roadmap that demonstrates a strong consideration for security features
• Public documentation of the project’s release versioning policy
• Easily understandable release process

Release and Update Process

Release Documentation

Provenance Artifacts

Artifacts included with each release:

• Checksums
• Checksum.sig/pem
• in-toto

User Instructions for Validating Artifact Signatures

Reporting Security Incidents

Security Reporting Method

Security Reporting Ownership

Reducing Vulnerabilities Through Development

Dependency Management

Security Champions

The following maintainers take a special interest in project security:

• Louis Jette
• Maria Ines Parnisari

Roadmapped Security Improvements

• Increase OpenSSF Best Practices badge level.
• Implement a singleflight CheckResolver to avoid concurrent evaluation of overlapping subproblems. #1301

Security Issue Resolution

Responsible Disclosure

OpenFGA vulnerability management is described in the official project security documentation SECURITY.md .

Incident Response

The OpenFGA maintainers bear the responsibility of monitoring and addressing reported vulnerabilities. Identified issues undergo prioritized triage, with immediate escalation upon confirmation. The triage process is conducted in private channels.

Adhering to the GitHub security advisory process, OpenFGA initiates the CVE (Common Vulnerabilities and Exposures) request upon issue identification. The resolution is developed in a private branch associated with the CVE.

Upon confirmation of the fix’s effectiveness, it is released through a new patch for each major supported version of OpenFGA.

The changelog will link to the CVE, which will describe the vulnerability and its mitigation. Any public announcements sent for these fixes will be linked to the release notes .

All OpenFGA security issues can be found on the Github advisories page .

Closed security issues and vulnerabilities

At the time of the joint assessment, OpenFGA listed closed issues under the security tab of the OpenFGA repo.

Hands-on assessment

The hands-on assessment is a lightweight review of the project’s internal security as well as the current recommendation configuration, deployment, and interaction with regard to security. Hands-on assessments are subject to security reviewer availability and expertise. They are not intended to serve as an audit or formal assessment and are no guarantee of the actual security of the project.

OpenFGA did not receive a hands-assessment from TAG-Security.

Roadmap

• Project Next Steps. The team has created a project under the OpenFGA organization to track remediations for the findings from this joint assessment. The required fixes are being undertaken by the project team.
• CNCF Requests. In the initial draft, please include whatever you believe the CNCF could assist with that would increase security of the ecosystem.

Appendix

• Known Issues Over Time. Past issues that have been publicly reported are listed under security issues tab in the repository. This is not a comprehensive list of all security issues.

Case Studies

The list of projects that utilize OpenFGA include Okta FGA, Twintag, Mapped, Procure Ai,Canonical (Juju & LFX), Wolt, Italarchivi, Read AI, Virtool, Configu, Fianu Labs, and ExcID.

Related Projects/Vendors

The list of related projects is available as a community resource