# Epic SI-248 - Scope API tokens to specific access levels

Project: Stream Integration Basics
Status: needs clarification
URL: https://community.openproject.org/work_packages/SI-248

## Detailed specification

### Motivation and background information

A specific use case for this are MCP clients or other AI agents, which may run on different trust levels. E.g. It might be fine to grant a SaaS LLM access to public data, but anything confidential and above is only allowed to be consumed by a privately hosted model on own hardware.

### Acceptance criteria

  - Resources (work packages, comments, meetings) have an associated access level
    
      - TODO: <mention class="mention" data-id="SI-249" data-type="work_package" data-text="###SI-249">###SI-249</mention>
    
      - TODO: <mention class="mention" data-id="SI-251" data-type="work_package" data-text="###SI-251">###SI-251</mention>
    
      - TODO: <mention class="mention" data-id="SI-250" data-type="work_package" data-text="###SI-250">###SI-250</mention>

  - Access tokens require certain scopes to read/write certain access levels
    
      - thus applications can be barred from access levels, by never granting them the corresponding scope

  - Admins can decide which scopes can be added to *personal* API tokens on a global level

### Out of scope

**Applying access levels to users**

The concept of access levels can also be used to restrict users from viewing certain data. This could either be enforced in addition to permissions or by preventing the inclusion of users in a project based on their access levels.

This Epic here is about limiting access for API clients based on the tokens they receive, not about limiting the access of the user represented by the token. However, both topics would probably share the same concept of access levels.

| Attribute | Value |
| --- | --- |
| Mockups |  |

## Planning

| Attribute | Value |
| --- | --- |
| Roadmap | 2027 Q1 |
| Date |  |
| Votes |  |
| List |  |
| Target versions |  |
| Relevance |  |

## Subitems

| ID | Type | Subject | Status | Assignee | Target versions | Project | Priority | Spent time |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| SI-249 | Open Point | Which access levels do exist? | new |  |  | Stream Integration Basics | Medium |  |
| SI-250 | Open Point | Do access levels form a hierarchy? | new |  |  | Stream Integration Basics | Medium |  |
| SI-251 | Open Point | How to assign access level to a resource? | new |  |  | Stream Integration Basics | Medium |  |
| SI-252 | Feature | Add scopes to personal API tokens | in specification |  |  | Stream Integration Basics | Medium |  |
| SI-253 | Feature | Restrict which scopes can be added to personal API tokens | in specification |  |  | Stream Integration Basics | Medium |  |
| SI-254 | Open Point | Glossary: How do we call things? | new |  |  | Stream Integration Basics | Medium |  |

## Bugs and other work packages

| ID | Type | Subject | Status | Assignee | Target versions | Module |
| --- | --- | --- | --- | --- | --- | --- |
| SI-254 | Open Point | Glossary: How do we call things? | new |  |  |  |
| SI-251 | Open Point | How to assign access level to a resource? | new |  |  |  |
| SI-250 | Open Point | Do access levels form a hierarchy? | new |  |  |  |
| SI-249 | Open Point | Which access levels do exist? | new |  |  |  |

## Details

| Attribute | Value |
| --- | --- |
| Module | API |
| Enterprise plan | Community (COM) |
| Labels |  |

## People

| Attribute | Value |
| --- | --- |
| Assignee |  |
| Accountable |  |
| Product manager | Nova Bonic |
| Feature Lead |  |
| Designer | Marc Alcobé |
| Developers | OP Dev Titan Team |
| Requested by |  |
| Interest by |  |

## Estimates and progress

| Attribute | Value |
| --- | --- |
| Work |  |

## Description

**As** an administrator,  
**I want to** add scopes to access tokens for limiting access to resources based on how confidential they are,  
**so that** I can hand out more access to applications with higher trust and less access to applications with lower trust.

#### Risks

*   Adding too much UX complexity for the problems we solve
    
    *   access levels have to be understood next to roles &amp; permissions
        
    *   proper access levels must be assigned to resources
        
*   Leaving restrictions based on access levels out (see &quot;out of scope&quot;) might make the feature harder to understand or behave surprisingly
