Agentology

Diagram of a stream of green and orange code blocks splitting into two paths, one billed at four Claude tokens and one at a single Claude token

Sep 5Link · 7m · engineering.atspotify.com ↗

Portal by Spotify cut my Claude Code token usage by 90%

Gitdailies — Pull Request Metrics & Advanced Alerts for GitHub

Aug 26Link · 4m · arxiv.org ↗

HANDBOOK.md: A Benchmark for Long-Context Agentic Instruction Following

Become a sponsor

HEADLONG on a beige CRT monitor

Aug 26Link · 12m · laude.org ↗

Headlong: a microharness for persistent agents

Aug 26Link · 17m · discoveredmaterials.com ↗

Material Discovery Bench

Red, orange and cream pixels scattered on deep maroon

Aug 26Link · 49m · danluu.com ↗

How does programming language affect token efficiency and correctness?

Red, orange and cream pixels scattered on deep maroon

Aug 25

The agent loop fits on a page; production does not

Red, orange and cream pixels scattered on deep maroon

Aug 24

Evals are the unit tests of agent systems

Red, orange and cream pixels scattered on deep maroon

Aug 23

Postmortems for something nobody typed

Black title card reading “OpenAI Agent Intrusion Technical Timeline” in monospace type

Aug 20Link · 51m · huggingface.co ↗

Anatomy of a frontier lab agent intrusion: the July 2026 timeline

“Agent autonomy, explained.” over ctrl and Z keycaps on grid paper

Aug 20Link · 7m · newsletter.posthog.com ↗

How much can you delegate to agents?

Aug 20Link · 19m · anthropic.com ↗

Patterns and problems in multiagent systems