Morning · Thursday, May 28

AI Brief

CLI proxy to cut LLM token costs 60-90%

  • 12 items
  • 2 axes
  • 3 sources
  • Generated 08:15 UTC

TL;DR

  1. 01
    github CLI proxy to cut LLM token costs 60-90%
  2. 02
    github Production-grade agent skills repo by Addy Osmani
  3. 03
    github Claude Code HUD plugin for context/tool visibility

Skills 9 items · Claude Code patterns worth stealing

  1. 01

    rtk-ai/rtk ★55,447

    github 1h ago score 9

    CLI proxy to cut LLM token costs 60-90%

    github.com/rtk-ai
  2. 02

    addyosmani/agent-skills ★46,569

    github 1h ago score 9

    Production-grade agent skills repo by Addy Osmani

    github.com/addyosmani
  3. 03

    jarrodwatts/claude-hud ★23,905

    github 5h ago score 9

    Claude Code HUD plugin for context/tool visibility

    github.com/jarrodwatts
  4. 04

    sqlite AGENTS.md

    rss/simonw 8h ago score 9

    SQLite AGENTS.md pattern ready to clone.

    simonwillison.net/2026
  5. 05

    CC-Wiki: Turn Claude Code sessions into a shareable knowledge base wiki

    hn 13h ago score 8

    Claude Code wiki tool, directly usable

    github.com/tejpalv
  6. 06

    gastownhall/beads ★24,154

    github 2h ago score 8

    Memory upgrade plugin for coding agents

    github.com/gastownhall
  7. 07

    obra/superpowers ★210,159

    github 8h ago score 8

    Adoptable agentic skills framework for Claude Code.

    github.com/obra
  8. 08

    Researcher "gave Claude Code 'ADHD' and it thinks 2x better now."

    hn 11h ago score 7

    Claude Code hack to improve reasoning, actionable

    thenewstack.io/claude-code-adhd
  9. 09

    Skill-as-Pseudocode: Refactoring Skill Libraries to Pseudocode for LLM Agents

    rss/arxiv-cs-cl 4h ago score 7

    Pseudocode skill libraries pattern directly applicable to Claude Code skills.

    arxiv.org/abs

General 3 items · Industry · models · policy

  1. 01
  2. 02

    Laguna M.1/XS.2 Technical Report

    rss/arxiv-cs-ai 4h ago score 7

    New coding-focused MoE model for agents

    arxiv.org/abs
  3. 03

    Coherence Collapse: Diagnosing Why Code Agents Fail After Reaching the Right Code

    rss/arxiv-cs-ai 4h ago score 7

    Diagnoses code agent failures, directly relevant to Claude Code usage

    arxiv.org/abs