A folder of 21 markdown files just hit #1 on GitHub trending. mattpocock/skills crossed 28,000+ stars on April 27, up from 18,700 two days earlier. No SDK. No framework. No application code. Each skill is a SKILL.md file under 200 lines, with YAML frontmatter that tells the
AI
-
METR Evals: Best Public Signal for AI Capability Progress
By
–
METR's evals are honestly the best public signal we have for actual capability gain right now, glad they're getting more airtime.
-
Inference Latency Optimization Gains Production Advantage
By
–
Wider also helps inference latency, this could end up more interesting for production than the CE win itself.
-
Agent-First Benchmarks: Evaluating Variable Step Protocols
By
–
Agent-first benchmarks are something! Curious what the protocol looks like for agents that take wildly different numbers of steps.
-
Industrial DevOps: Continuous Manufacturing Deployment Strategy
By
–
This approach applies CI/CD principles to manufacturing through Industrial DevOps. Individual changes instead of accumulated batch deployments.
-
Building AI Agents with Skills vs without Skills
By
–
Building AI Agents with Skills vs without Skills https://t.co/KZR4xoFppY pic.twitter.com/lI8nW35qB8
— Shubham Saboo (@Saboo_Shubham_) 27 avril 2026Building AI Agents with Skills vs without Skills
-
Compute Wars Intensifying: A Critical Moment in AI Hardware
By
–
Compute wars intensifying. What a time to be alive haha.
-
Evan Spiegel on Stories: Innovation Through Refusing User Demands
By
–
.@evanspiegel shares the story behind Stories, and how it came to be by refusing to build what users wanted https://t.co/koxH5vJWOZ pic.twitter.com/dsd6Mkb0xA
— Lenny Rachitsky (@lennysan) 27 avril 2026.
@evanspiegel shares the story behind Stories, and how it came to be by refusing to build what users wanted -
AI Often Judged on Outdated Models, Skewing Research Results
By
–
This points to a broader issue where AI is often being judged based on extremely outdated models. Any time someone sends me a study showing AI failing at something, the first thing I do is look at the models they used… more often than not, they’re extremely outdated. We need