[Initiative]: TOC AI Cloud-Native Standards Checklist

[joshhalley]

Name

Kubernetes Agentic Standards Checklist Creation

Short description

Creation of a best practice recommendation checklist for Agentic AI micro-services

Responsible group

TOC

Does the initiative belong to a subproject?

Yes

Subproject name

Kubernetes Agentic Standards Checklist

Primary contact

joshhalley

Additional contacts

Nina Polshakova

Initiative description

The objective of this initiative is to provide a concise checklist in under 10 pages associated with best practices which have been learned around the deployment and creation of micro-services in the domain of AI Agentic containers.

Key topics are based of the pre-work that the team pursued in the below working document:

working document

Deliverable(s) or exit criteria

The objective is to create concise and easy to understand checklist covering the topics described in the Document Link that can support individuals and teams that are developing agents to consider important topics, in domains of security and observability.

The objective from the team is to have the document concluded by the end of July 2025.

[leonardpahlke]

This reads a bit like a talk proposal 😄 could be a great blog too!

[angellk]

+1 @leonardpahlke -- looks like it could also be a short, concise Initiative that the TOC is looking for (versus the longer whitepapers)

okay to open vote @riaankleinhans

[riaankleinhans]

This initiative has been approved by the TOC and is ready to be worked on with the appropriate TAG and TOC liaison.

[joshhalley]

Kubernetes Agentic AI Standards:

Document Link:
https://docs.google.com/document/d/1pH7b2IKNg6lDcHNHZJoDdWNoe8jvp3Kd0Aoy7vTdAOA/edit?tab=t.0

Attendees:
Deep Patel
Claudia Misale
Victor Lu

cc:
Ricardo Aravena
Nina Polshakova
Vincent Caldeira
Nimisha Mehta

Meeting Minutes - Tuesday July the 10th - 08:00 PT:

Hi Team,

please find the minutes from todays session:

• Overview was shared for the format of the document

  • Max 10 pages - short abstract followed by checklist

• Described how supporting or reference documentation associated with checklist should be handled

  • Links for these documents to be included at the bottom of each sections checklist

• How do we handle disclaimers for fair use, to indicate that this document does not mean that users are not at risk of security breach etc...

  • Disclaimer can be included in document, however there is an expectation that something exists already (boilerplate) from the CNCF

• Deep volunteered to support Vincent on Governance section

  • Deep to reach out to Vincent

• Victor indicated that a section on trust is missing and should be added

  • Added to the governance section of the document

• Topic of up and coming / future research was discussed

  • For future topics, limited abstracts can be added at the end of each section, however given the pace of change in this domain, it is expected that change revision is high

• Meeting Schedule Tue / Thu at 08:00 AM Pacific

  • Josh to reach out to Riaan to make the meeting series less restrictive and get the Tuesday slot on the schedule

In case any topics were missed or misrepresented let me know.

Kind Regards,

Josh

[joshhalley]

Kubernetes Agentic AI Standards:

Document Link:
https://docs.google.com/document/d/1pH7b2IKNg6lDcHNHZJoDdWNoe8jvp3Kd0Aoy7vTdAOA/edit?tab=t.0

Attendees:
Nimisha Mehta
Claudia Misale
Victor Lu
Nina Polshkova

cc:
Ricardo Aravena
Deep Patel
Vincent Caldeira

Meeting Minutes - Tuesday July the 15th - 08:00 PT:

Hi Team,

please find the minutes from the session:

• Discussed the General section, indicating that it should set the stage and cross reference other links to avoid base jumping into content which is too heavy
• Low Level Hardware topics (i.e. Metrics for NVIDIA vs AMD etc...) will be called out as out of scope, as this domain is quite exhausive and goes well beyond agents
• Various topics converge across domains in terms of trust and security (such as accounting for audit trails, arguably could also be a observability metric), team agreed should work with the current format and we can always reshuffle the final to make it more readable

Target delivery remains end of July

Meeting Schedule Tue / Thu at 08:00 AM Pacific, until the end of the month

In case any topics were missed or misrepresented let me know.

Kind Regards,

Josh