Building a way to see what AI coding agents did, issue by issue.
On a team it's hard to tell:
- how many runs went into an issue
- what happened in PR, CI and review after
Issue → Run → PR → CI → review, on one timeline.
Teams using AI coding: what's hardest to see?
A key perspective for building harnessforce.
I finally got around to reading "Issue tracking is dead" from Linear.
And yes, issues are breaking down more and more in my workflow. The speed of development and the productivity LLMs bring are dismantling what we used to take for granted.
Which is exactly why it's time to reinvent it.
link in reply 👇
5/5
Learned:
- ClickHouse errors weren't leaving error codes in the logs, so I had to query it directly to debug
- The Terraform ClickHouse Cloud provider didn't support my planned setup (tier, password handling), so I redesigned around its constraints
- Removing Backlog exposed an old asymmetric bug in the progress calculation. Fixed.
1/5
Day of #buildinpublic
Reworked the information architecture of every screen in harnessforce, our AI-native project management platform in development.
Went through ~700 Storybook patterns and 20 live screens, and analyzed 297 PRs merged in the last 3 months (96% co-authored with AI, 86% merged without review).
4/5
terraform-infra (production infra in Terraform):
- Declared the https://t.co/L1gdBdTpEA DNS zone as code
- Added a test that compares responses from the two nameservers
- Documented the plan to manage telemetry in ClickHouse Cloud
@JaiOnX20@X Hey Jai! 👋 I’m a software developer from Japan, working on my own projects with AI coding agents. Great to meet others into AI and building things! What are you working on today?
Just signed up for ClickHouse for the first time. I can't contain my excitement! 🚀
Going to use it as the telemetry backend for HarnessForce and make it blazing fast!
@IndieDevFire Really agree with that.
I'm aiming to release next month, so I'll let you know when it's out! The idea is basically "Datadog for the LLM era".
It connects to GitHub to make black-box LLM-driven development visible.
Building a way to see what AI coding agents did, issue by issue.
On a team it's hard to tell:
- how many runs went into an issue
- what happened in PR, CI and review after
Issue → Run → PR → CI → review, on one timeline.
Teams using AI coding: what's hardest to see?
@jimmy_dj_c Respect the honesty, Jimmy! Finding the right problem is tough. I’m building a tool to track AI coding agents’ work. Let’s connect and learn together—feel free to follow along! Rooting for your first paying customer 🙌
@Devvora15 Hey Dev! 👋 I’m a developer from Japan building web apps with TypeScript, Next.js, and AI coding agents. Would love to connect with fellow builders! Feel free to follow me—let’s share ideas and support each other 🤝
@abinav_exe@X Hey Abinav! 👋 I’m a developer from Japan working with TypeScript, Next.js, and AI coding agents. Love building projects and exchanging ideas with fellow devs. Let’s connect! 🤝
@stackway24 Hey gab, great to connect! 🤝
I’m building with AI and sharing my indie dev journey too. Followed—looking forward to learning from each other and seeing what we build!
🚀 I’d love to connect with more people in tech!
💻 Developers, founders, designers, AI builders, and anyone interested in technology are all welcome.
🛠️ I’ll be sharing what I’m building, the challenges I’m tackling, and my experiments with indie hacking, AI, and web development.
🤝 Let’s learn, build, and grow together!
#IndieHackers #AI #WebDevelopment #Startups #Connect #TechTwitter