I'm looking for recommendations on AI tools that can help streamline my workflow and support my development team. Currently, I'm using Kubernetes MCP and Claude Code, but I'm interested in discovering more tools to ease my workload, especially as I face a lot of tech debt. Any suggestions for new tools to experiment with would be greatly appreciated!
7 Answers
Don’t forget about all the command-line tools! I use the Atlassian CLI for ticket creation, and Grafana/Loki CLI for querying logs. They really enhance productivity with direct access.
K8sGPT is definitely a tool worth exploring for managing cluster issues. It really cuts down the time spent manually checking pod logs, helping devs focus more on building features instead of firefighting daily operational issues.
For heavy tech debt sessions, glm-5.1 works wonders inside Claude Code. It cuts down on costs while grinding through all that legacy code—definitely a lifesaver! Keep at it!.
We're using Claude Code alongside detailed .md files that explain our architectural decisions and system connectivity. It acts as a knowledge base for our devs. I'm also utilizing it for planning phases, like when we want to port old Lambdas to CDK. It's great for utilizing downtime during less coding-heavy days in meetings.
I've found the plugins within Claude for gcloud observability to be fantastic for troubleshooting. They allow querying across multiple services which is super handy when you're digging through logs.
I actually built Delimit to handle context switching between Claude Code, Codex, and Gemini CLI. It's a server plus CLI that keeps everything consistent, and it can even catch breaking API changes before they go out. I'm also looking into incorporating third-party tools next!
I've had great success with Gemini! It's been really helpful for creating study guides for Kubernetes and troubleshooting common issues, like figuring out what to do when a pod gets stuck in a pending state. For coding tasks, I rely on Claude.ai; it's great for that.
Exactly! It's surprising how many people overlook simple errors when interacting with LLMs. Sometimes, you just need to know the basics.

That sounds smart! Have you measured how much time it saves or the cost associated with running those features?