Coding agents can now analyze 80 files and run a full test suite in minutes. So why are we still reading 1,800 words of generated Markdown to figure out what they actually did?
Today I am proposing a new standard to fix this: the Present Specification.
But the terminal is just the starting point. Because the specification is decoupled from the UI layer, the roadmap goes further:
• v0.1: Terminal and Pi
• v0.2: Web renderers (React, MCP Apps)
• v0.3: Voice interfaces and mobile displays
Agents should not write reports. They should present results.
The zero-dependency specification, the schema, and the reference implementations are all open source. I am looking for feedback on the schema design and possibly collaborators.
Code and spec: github.com/sjespersen/agent-…
Low Quality Image Placeholders: guypo.com/introducing-lqip-l… #webperfdays
Sebastian retweeted
List of the points made in the 3rd Party session #webperfdays
Sebastian retweeted
Choosing the topics for the break out sessions #webperfdays
Sebastian retweeted
Great crowd at #webperfdays event. Fantastic follow-up to #velocityconf.
If you do ops. Go read @allspaw's thesis. I'll wait. bit.ly/AllspawThesis
Sebastian retweeted
Slides for @bluesmoon and my Forensics Tools for In-depth Performance Investigations #VelocityConf slideshare.net/mobile/nicjan…
Sebastian retweeted
Leading cause of lost productivity - the incredible Norse Live Attack Map... It's mesmerizing! map.ipviking.com/ #infosec #security
Just got to #webperfdays Barcelona. Perfect companion to #velocityconf
Sebastian retweeted
.@aerotwist is on the livestream at the Dev Summit now for his talk "The Applied Science of Runtime Performance" developer.chrome.com/devsumm…
Sebastian retweeted
CSS Triggers csstriggers.com/ tells you what property changes will cause layouts, paints and composites
















