What I did today:
- Refactored opencode-feishu-notifier; it now sends notifications to users in a defined way.
- Continued letting AI write legionmind-github-bridge. I started using opencode’s multi-agent mode. It launched 5 agents to address 5 issues, chugging away for 2 hours and consuming all 5 hours of my codex tokens.
- Today, one node in the SG cluster died. I checked the logs and found it was being attacked by continuous SSH attempts. That’s not good. After a quick assessment, a few possible directions:
- Disable password authentication
- Disable the SSH daemon facing the entire network
- Move the cluster behind a NAT
- Handled some miscellaneous tasks. ZL is coming to Suzhou next week. I spent some time making arrangements, but it didn’t go smoothly. I don’t plan to invest more energy in this.
Thoughts:
At this stage, I can only manage 2-3 things simultaneously, including development work, daily schedules, thinking, and output. Beyond that, I struggle to keep up and easily feel exhausted. This is despite my efforts to delegate work to AI agents as much as possible. Therefore, I think there are two areas for improvement:
- For coding tasks, I should improve the autonomy of agents as much as possible, with the following optimization goals:
- Minimize interruptions to me
- Maximize the amount of work they do
- Improve the reliability of their work as much as possible
- For myself, I also need to improve:
- Manage my mental energy better so I don’t get tired quickly
- Enhance my ability to work in multiple different contexts simultaneously, without losing track of tasks or forgetting details, and with progress management
Based on the above, I think tomorrow I can try two approaches:
- Design a multi-agent template for legionmind and experiment with it on a coding task in yuan using opencode
- Continue logging work to explore a method for managing mental energy and context
- For coding tasks, I should improve the autonomy of agents as much as possible, with the following optimization goals:
What I plan to do tomorrow?
- As mentioned above, set up a multi-agent experiment
- Continue working on legionmind-github-bridge
- If time permits, work on cluster security
—
Overall, my main focus now is to use AI to scale myself, and then try to scale others.
2026 Work Log
Browse the 2026 work log day by day.
13 则 第 3 / 3 页