New idea in testing: Agent skill that helps you set up a home lab docker deployment that periodically pulls its config from a git repo. Agent sets up framework and service configuration you review and production server pulls with no direct agent access. Want to add or configure new services? Make requests and edits on your local device and they deploy server side with minimal editing on server. Server stays lean and deployments are all version controlled. Will share code after more testing in action.

The open source tool set for running local agents has really improved. Setting up Hermes with access to tools like Hindsight has been a game changer. Also a great feedback loop in that the improved agent quality makes managing self-hosted tools easier and then makes the quality of the agent interaction stronger. Getting more fun by the day!

Lounge life

I’m not even a NASCAR fan, but Kyle Busch… damn.

American Airlines really has pretty garbage service for a top airline.

That feeling when you get banned for marketing/advertising after sharing open source code to try to help teachers. I’d be laughing if it wasn’t so sad and ironic.

Had some fun playing with Claude design and turned it into something functional. Built this to help my wife with class assignments they have to do every year at their elementary school. Commercial tools exist, but most schools don’t have them. Hopefully this helps some other teachers out there!

GitHub

GitHub - armstrys/class-list-builder: A simple web app to optimize class lists based on student scores and needs.

A simple web app to optimize class lists based on student scores and needs. - armstrys/class-list-builder

Any folks interested in NCAA basketball on nostr? Working on a long-term project and considering putting out a daily digest. What would be interesting? I have a ratings method I developed… would you want to see: - predictions for key games? - betting insights? - not necessarily recs - Short blurbs on relevant news?

As a man I didn’t fully understand what an ick was until I saw the Seattle Seahawks trophy presentation happen without the team on stage. Especially gross bringing the QB on stage with the MVP for no apparent reason

Currently trying to implement a project team in openclaw… it’s not there yet, but it feels really close. Would be curious if others have tried this. MiniMax may not be smart enough to pull it off sustainably. - 4 agents - Project manager - Full stack developer - QA / Verification - Research and resource investigation - Agent to agent communication enabled in config - Each agent has independent communication channel - Convex database in a convex folder with following features: - Task management (create, assign, update status, block/unblock) - Message logging - SearXNG installed with docker for local search - Cron jobs for each agent to check messages, tasks, and blockers every 45 minutes - Each agent has a WORKING.md file where they log their progress, blockers, and key findings. This file is also used for other agents to check if their teammates are blocked or need help. - Cron job for project manager to check all agents' WORKING.md files every 45 minutes to identify blockers and prioritize unblocking as the number 1 task and then do his own work. - Daily cron job for project manager to summarize all work and run stand up - Daily cron job to check system health - Daily cron to review project materials / scoping / planning in the projects folder Steps: 1. Get agents running. 2. Make sure agents can communicate with each other 3. Get the project manager to put together a clear specification for how communication will work with convex 4. Have full stack developer implement the specification and create a skill describing how the database is used 5. Have full stack developer set up SearXNG search engine and create a skill describing how to use it 6. Have QA, Researcher, and PM test the tools 7. Get the PM to work with you on scoping a long-term projects and assigning initial tasks. 8. See if it works - check and adjust as needed.

I was totally confused by the association of clawdbot/moltbot/openclaw to Mac mini purchases. It’s not free. It’s not sovereign. It’s definitely not secure. It’s flexible. And it’s going to be a gateway to people running their own sovereign services. If I could nip-46 the permissions for everything my agent is working on I would feel a lot more comfortable… most people in the US don’t even know that’s possible. Let’s make sure that flood of Mac Mini purchases gets put to good use. @jb55 @jex0 ⚡ @npub1xy39...ewcs @npub1m3fy...7j6d