gjkim042

5 ms·

1.
by gjkim042 7mo ago
> I guess I could instruct the agent to build a plan and to post the plan in the issue and then iterate that with written comments in the issue. Is that how you run it? Exactly. I simply just create an issue if there's a simple bug
2.
by gjkim042 7mo ago
It was the decision to simplify the architecture. And actually workspace(github) is a persistent layer for kelos. You can create a new branch and new PR and updage it on next iteration based on comments and existing patch.
3.
Show HN: Kelos – Run Claude —dangerously-skip-permissions on Kubernetes(github.com)
2 points by gjkim042 7mo ago|1 comments
4.
by gjkim042 7mo ago
That makes a lot of sense. Thanks for the feedback.
5.
by gjkim042 7mo ago
Get the idea. I was wondering what’s the real scenario that needs agent chaining? I implemented the task chaining using “dependsOn” field, but there’s no real usage now. Can we just write a pipeline prompt using the numbering? 1. Do this, 2
6.
by gjkim042 7mo ago
Thanks, Actually, Kelos captures structured outputs after each coding task compeletes: the final branch, the created PR link if exists, input/output tokens, and so on. And coding agent’s logs would remain after finished as a container
7.
Show HN: Kelos – Define your AI coding agent workflow as YAML on Kubernetes(github.com)
4 points by gjkim042 7mo ago|5 comments
8.
Show HN: I run a team of AI agents on my Kubernetes cluster(github.com)
2 points by gjkim042 7mo ago|0 comments
9.
by gjkim042 7mo ago
about the adoption friction, You can easily use it by kind create cluster + axon install. Thanks for your interest. (Sorry for the late interaction, I thought the post was just abondoned.)
10.
by gjkim042 7mo ago
about the agent container interface, It’s mostly about the inputs not the outputs. (How to inject prompts, context, skills…) However, axon is trying to provide deterministic outputs such branch, PR link, etc, by executing a post script.
11.
by gjkim042 7mo ago
About handling secret/credential scoping, there’s todo issue to reference kubernetes secret from mcp config. https://github.com/axon-core/axon/issues/313
12.
Show HN: Axon – A Kubernetes-native framework for AI coding agents(github.com)
1 points by gjkim042 7mo ago|4 comments
13.
Show HN: Axon – Safely run claude --dangerously-skip-permissions on Kubernetes(github.com)
1 points by gjkim042 8mo ago|0 comments
14.
Show HN: Axon – Let coding agents develop their own framework on Kubernetes(github.com)
3 points by gjkim042 8mo ago|1 comments
15.
by gjkim042 8mo ago
Hello HN, I built Axon because I wanted to run Claude Code fully autonomously but didn't want to give it free reign over my laptop. The insight was simple: Kubernetes already solves the hard problems. Isolation? Pods. Ephemeral environ
16.
Show HN: Axon – Run autonomous coding agents(Claude, Codex) safely on Kubernetes(github.com)
4 points by gjkim042 8mo ago|1 comments