mirror of
https://github.com/kanidm/kanidm.git
synced 2025-02-23 20:47:01 +01:00
Update readme
This commit is contained in:
parent
6ba0f6fccb
commit
4934894666
63
README.md
63
README.md
|
@ -9,45 +9,52 @@ Kanidm is an identity management platform written in rust. Our goals are:
|
||||||
|
|
||||||
* Modern identity management platform
|
* Modern identity management platform
|
||||||
* Simple to deploy and integrate with
|
* Simple to deploy and integrate with
|
||||||
* extensible
|
* Extensible for various needs
|
||||||
* correct
|
* Correct and secure behaviour by default
|
||||||
|
|
||||||
|
Today the project is still under heavy development to achieve these goals - we don't expect a fully
|
||||||
|
functional release before early 2020.
|
||||||
|
|
||||||
## Code of Conduct
|
## Code of Conduct
|
||||||
|
|
||||||
See CODE_OF_CONDUCT.md
|
See [CODE_OF_CONDUCT.md]
|
||||||
|
|
||||||
## Examples
|
[CODE_OF_CONDUCT.md]: https://github.com/Firstyear/kanidm/blob/master/CODE_OF_CONDUCT.md
|
||||||
|
|
||||||
## MVP features
|
## Some key ideas
|
||||||
|
|
||||||
* Pam/nsswitch clients (with offline auth, and local totp)
|
* All people should be respected and able to be respresented securely.
|
||||||
* CLI for admin
|
* Devices represent users and their identities - they are part of the authentication.
|
||||||
|
* Human error occurs - we should be designed to minimise human mistakes and empower people.
|
||||||
|
* The system should be easy to understand and reason about for users and admins.
|
||||||
|
|
||||||
|
## Quick start
|
||||||
|
|
||||||
|
Details to come ...
|
||||||
|
|
||||||
|
## Implemented/Planned features
|
||||||
|
|
||||||
|
* RBAC design
|
||||||
|
* SSH key distribution for servers
|
||||||
|
* Pam/nsswitch clients (with limited offline auth)
|
||||||
|
* Sudo rule distribution via nsswitch
|
||||||
|
* CLI and WebUI for administration
|
||||||
* OIDC/Oauth
|
* OIDC/Oauth
|
||||||
* SSH key distribution
|
* Claims (limited by time and credential scope)
|
||||||
* MFA (TOTP)
|
* MFA (Webauthn, TOTP)
|
||||||
* In memory read cache (cow)
|
* Highly concurrent desgin (MVCC, COW)
|
||||||
* backup/restore
|
* Replication (async multiple active write servers, read only servers)
|
||||||
|
|
||||||
## Planned features
|
|
||||||
|
|
||||||
* Replicated database backend (389-ds, couchdb, or custom repl proto)
|
|
||||||
* SAML
|
|
||||||
* Read Only Replicas
|
|
||||||
* Certificate distribution?
|
|
||||||
* Web UI for admin
|
|
||||||
* Account impersonation
|
* Account impersonation
|
||||||
* Webauthn
|
* RADIUS integration
|
||||||
* Sudo rule distribution via nsswitch?
|
* Self service UI with wifi enrollment, claim management and more.
|
||||||
|
* Synchronisation to other IDM services
|
||||||
|
|
||||||
## Features we want to avoid
|
## Features we want to avoid
|
||||||
|
|
||||||
* Audit: This is better solved by ...
|
* Auditing: This is better solved by SIEM software, so we should generate data they can consume.
|
||||||
* Fully synchronous behaviour: ...
|
* Fully synchronous behaviour: This is slow.
|
||||||
* Generic database: ... (max db size etc)
|
* Generic database: We don't want to be another NoSQL database, we want to be an IDM solution.
|
||||||
* Being LDAP: ...
|
* Being LDAP/GSSAPI/Kerberos: These are all legacy protocols that are hard to use and confine our thinking - we should avoid "being like them".
|
||||||
* GSSAPI/Kerberos
|
|
||||||
|
|
||||||
## More?
|
|
||||||
|
|
||||||
## Get involved
|
## Get involved
|
||||||
|
|
||||||
|
|
Loading…
Reference in a new issue