Saw a few posts this week about tracking Claude usage, and I get it, the limit anxiety is real. But the thing that actually fixed it for me wasn't watching the meter closer. It was changing how I work so I stop burning through it. Four things that made the biggest difference: 1. Plan before it builds anything. Most wasted tokens come from one long session piling up dead ends. I make it lay out the plan in a few bullets first, then approve it. A minute of planning saves a pile of failed retries. 2. Push heavy reading into a subagent. Instead of dumping a huge file or a load of research into my main chat, I send it to a subagent. It reads the 50k tokens and hands me back a 2k summary. My main context stays clean and cheap. 3. Start fresh before the context fills up. Quality drops as the window fills, and a bloated session costs more per message. Around 70-80% full I save a short checkpoint, clear, and pick up from there. Feels backwards but it's faster and cheaper. 4. Keep instructions in files, not in every prompt. Anything reusable lives in a file the agent loads only when it needs it, instead of me re-explaining it every session. Net effect, I almost never hit the wall now, and the meter went from something I anxiously refresh to a signal that a session is getting bloated. What eats your limit the fastest? Curious if it's research, long builds, or something else.

🔥 Day 3 of the AIS 7 Day Challenge Used Nate's Skill Builder for Claude Code to build a chatbot that logs my daily, weekly and monthly execution reviews from The Less Paradox (a transformational seminar by Lakers Kokmaiya at the Pillars Institute). Optimisation from today: split the planning questions and review questions into two separate groups — way cleaner flow, and the bot stopped mixing forward-looking prompts with reflection prompts. Biggest takeaway so far: the Skill Builder makes it stupidly fast to turn a personal framework into a working tool. On to Day 4 💪

So this is the most interesting project I have done so far for sure... and I am already imgining multiple case scenarios in which this might be very useful for me, especially with complex workflows involving external databases, API's, and other stuff like that. One of the skills in this AI OS is to simply give me a rundown of the unread e-mails I had from the day I am requesting that skill to be used (i.e. the last 24 hours), and I can also create a to-do list on ClickUp for some e-mails that need urgent responses and rank based on priority. Very fun project, and this AI OS may have a huge amount of potential to be better. Loom video: https://www.loom.com/share/976fcdd8290e40a4bd2a87a27664e64c

#AIS Challenge I have created a Scheduled Task to check Lowest Flight Ticket Prices until next run From Toronto to Melbourne on Every Sunday at 3:00 PM on Weekly Basis. For Loop, I set a reminder to drink water and look away from screen for 10 seconds, and it will remind me after every 10 minutes. Also created one time reminder to take screenshot after 60 seconds for demo. I found Scheduled task more helpful, because It will let me schedule my skills to run by its own that I otherwise run manually. Even where I used for Day 6, is gonna be helpful to track ticket prices for a week. For loop, If You start learning new things, you can use loop to remind you something which you could forget otherwise. The Self-healing is a thing that I always praise these AI Agents. It makes the learning AI a piece of cake for People who think learning AI is difficult. Tools Used: Claude Code Pro, VS Code, Firecrawl

How are you guys getting the most out of Fable 5? Any tips? Thanks.