Quick end-of-week update, starting with a name change. Claude Analytics is now CheckYourAgent. I'm adding support for Codex next, and likely more agents after that, so a Claude-only name stopped making sense. Same tool, same repo, broader scope. What shipped this week: Some new features: - Team support. It's no longer a solo tool; you can pool a whole team's Claude Code usage and read the analytics in one place. Sharing has privacy levels you set to match your team's policy. - Price history. Cost now uses the token price at the time of each session, not just today's prices. - Sturdier imports. A batch of fixes and performance work so large imports don't fall over. A UI pass: - Sleeker look overall. - Loading states everywhere they were missing. A real difference for heavy users with lots of big projects. - First pass at making the busier views calmer and less overwhelming. - Privacy mode to hide session details, handy for sharing screenshots. WIP: - Keeping a local archive of sessions as Claude prunes older data. Next up (hopefully within two weeks): - Feature parity between team and individual views. - OpenTelemetry for a live feed, alongside manual exports. - Codex support: the tricky one, since I want it at full feature parity, not bolted on. Still local-first and open source. Nothing leaves your machine unless you send it yourself.

Hello everyone! Happy to share a tool I’m actively developing. I know there are already several great tools for tracking Claude token spend locally, but I felt there was a lot more we could do with the data Claude Code exposes on the machine. The cost view is still the star of the show for most people. But for anyone who wants to dig deeper into their usage, understand what is actually driving costs, how their agents operate, and how all of that changes over time, I think this tool can help. I’m using common, and some less common, knowledge discovery techniques such as sequence mining and subgroup discovery to build features like session forensics, context economics, usage mind maps, and more. All while trying to keep the data visualisation useful, clear, and pleasant to look at. It’s still a WIP, so some parts need refinement, but it’s open source and I’d be very happy to get feedback here or on the GitHub repo: https://github.com/TarekAwwad/authrty-claude-code-analytics Also, here is the project landing page for those who prefer nicer UIs: https://checkyouragent.dev/

No fancy setups, no theory. Just the small workflow that actually earned its keep. Mine is a simple daily email digest that pulls from my inbox and calendar and sends me a 5 line morning brief. Sounds basic but it saves me the first 30 mins of every day. Drop yours below. Even if it feels too simple, that's usually the stuff that helps someone else the most. What it does, what tool you built it with, and roughly how much time it saves you. Let's make a thread people can actually steal ideas from.

Hello guys, The last two days I have faced an issue I never thought possible. I have the Max plan (the $200 version, so the most expensive one). YESTERDAY, when I got to the office I had 46% of my 5 hour window already consumed although it had barely began. But I though it should be from my morning reports. However, these never used more than 5-10% of the 5 hour limit before. But this time I discovered there were 2 sessions running a total of about 65 tasks that appeared to have been running for over 60 hours. When I checked, those had been finished shortly after they had been launched, because they were short tasks. I asked my claude code agents to close every task and all my agents confirmed having closed the tasks they had been running. But the app still showed them running, so I closed them manually. Then I cleared the contexts and went on with my work normally, as if nothing had happened. TODAY: I got to the office about 30 minutes earlier, so the morning reports hadn't triggered yet. However, I appeared to have used 100% of my 5 hour limit that had just started. Has anyone else encountered this kind of issue? How can we solve this? NOTE: I asked support and it asked me to close manually all tasks and delete all sessions. Honestly that damages much of my work, but I still did this for most sessions (and discovered they also had running tasks although those sessions had been closed in forever - some of them were months old).

My entire life and business now live in one second brain, and Claude Fable understands it better than I do. In this video I walk through my full Claude Fable AI operating system using the four Cs framework: context, connections, capabilities, and cadence. You'll see exactly how my files and folders are set up, how I keep improving it every day, and the usage tips I rely on to get more out of it.