I'm working on putting together a solid definition for our agents that I can store in Git. My plan is to specify the role of each agent (like SRE, FinOps, etc.), the functions I expect them to perform (such as Infra PR review or triage alerts), and the systems I want them connected to (like GitHub, Jira, AWS, etc.). I've already drafted a structure that looks like this:
```yaml
agent:
name: Infra Reviewer
role_guid: "SRE Specialist"
connectors:
- connector: "github-prod"
type: github
config:
repos:
- org/repo-one
- org/repo-two
- connector: "aws-main"
type: aws
config:
region: us-east-1
services:
- rds
- ecs
- connector: "jira-board"
type: jira
config:
plugin: "Jira"
functions:
- "Triage Alerts"
- "PR Reviewer"
```
Once I finalize the definition, I'd like to integrate it into a GitOps workflow to keep agent configurations aligned. I'd love to hear your thoughts on whether this makes sense or if there's a better way to approach it!
1 Answer
One suggestion might be to use markdown files for agents along with a directory structure for their skills and specs, rather than just relying on Git. Depending on how often you plan to change these definitions, Git might not be the best fit for this task. It's worth considering how frequently updates will need to happen.
Thanks for your input! I think having a clear definition that's versioned would help maintain consistency across the team, and it would also make it easier for anyone to add more agents to the config. It's a similar approach to how we manage our infrastructure files in Git.