Seroter's Reading List: CLIs Save Tokens, Agent Harness Explained, When to Use Diffusion LLMs
rseroter · x · 2026-10-24
Richard Seroter's daily reading roundup highlights: a post showing CLIs are the most token-efficient way to equip agents; research on when developers accept proactive AI suggestions (only at natural work boundaries); a getting-started guide for the Google Cloud CLI Remote MCP Server; a crash course on what an AI "harness" actually is; and a DiffusionGemma vs Gemma 4 comparison on when diffusion LLMs make sense. Also covered: slow software is a design problem, not a language problem, and pushback on Rust overhype.
More from coding & agent
- Mapping Economic Theory With Agentic AI: LLM Agents Turn Theory Papers Into Graphs — mattbeane · 2026-10-24
- Claude launches Duet proactive agent preview; user shares self-hosted OpenClaw alternative on VPS + Slack — heyneighbor · 2026-10-24
- Coding agent cost breakdown: Claude Code tops Kotlin benchmark at 85.7% but burns 3.6x fewer tokens than Codex — Evening_Hawk_7470 · 2026-10-24
- Proof of Agent site logs 40k bots vs 4k humans in 24 hours — adamamcbride · 2026-10-24
- Vercel launches terminal-based budgets: cap team spend and auto-pause production at 100% — cramforce · 2026-10-24
- humans& ships collaborator model Ampersand, social product Boba, and open-sources 10% FTL decode speedup — ericzelikman · 2026-10-24