Skip to content

DEV Community

shimo4228

I run AI agents and publish everything I learn, failures included, from Claude Code harness engineering to an experiment running fully local on one M1 Mac.

Kobe, Japan Joined on Feb 14, 2026 https://orcid.org/0009-0002-6168-4162

Give Claude Code a Second Harness

#claudecode #anthropic #contextengineering #aiagents

8 min read

Want to connect with shimo4228?

Create an account to connect with shimo4228. You can also sign in below to proceed if you already have an account.

Already have an account? Sign in

Banned from Wikidata Overnight — I Believed Every Edit Was Compliant, but All 109 Items Were Deleted as Promotion

#wikidata #postmortem #geo #claudecode

13 min read

What Humans Should Approve Is Intent, Not the Diff — A Decision Table for Agent Approval Gates

#claudecode #aiagents #devops #ai

18 min read

Opus 5 Changed How Rules Should Be Written — Audit Yours

#claudecode #contextengineering #promptengineering #anthropic

12 min read

claude-security Beta: Cost, Output, Accuracy — 189 Agents, 2 Hours, Measured

#claudecode #anthropic #security #ai

11 min read

herdr, a tmux for AI Agents — Until the Editor Disappeared

#claudecode #terminal #productivity #ai

18 min read

What Do My AI Agent's Logs Look Like in OpenTelemetry?

#opentelemetry #observability #jaeger #llm

15 min read

Fault Injection TDD Found 3 Silent Failures in My LLM Agent

#testing #llm #pytest #agents

12 min read

LLM-as-Judge Shouldn't Aggregate Scores: Binary Checks as Evidence, One Holistic Verdict

#llm #promptengineering #evaluation #claudecode

12 min read

Building an Autonomous Agent on an M1 Mac, by Choice

#discuss #ollama #llm #agents

6 min read

Why Did My Agent Decide That? 3 Observability Patterns

#observability #agents #claudecode #skills

13 min read

Claude Code from iPhone: Plugging 3 Holes in Remote Control

#claudecode #tmux #mobile #github

14 min read

Linters Can't Measure a Non-Code Asset's Value — I Built an LLM Stocktake for It

#claudecode #ai #devtools #githubactions

9 min read

I Built a Skill for Easy Codex Reviews from Claude Code

#codereview #codex #claudecode #skills

7 min read

Between the Workflow and ReAct Quadrants: How Phase Decides Skill Design

#discuss #ai #programming

10 min read

Is ReAct Needed in Production? — Separating Design and Operation Phases

#discuss #ai #llm #architecture

10 min read

(3) The LLM Workflow Quadrant Is Missing from Our Vocabulary

#discuss #ai #llm #architecture

11 min read

Where ReAct Agents Are Actually Needed in Business

#discuss #ai #programming

14 min read

Organic Growth and Content Integrity in an AI Writing Team

#discuss #ai #claudecode #writing

6 min read

Building Zed as an Observation Window for Claude Code — Japanese Typography with IBM Plex

#zed #ai #font #typography

10 min read

AI Agent Black Boxes Have Two Layers — Technical Limits and Business Incentives

#discuss #ai #governance

8 min read

Can You Trace the Cause After an Incident?

#discuss #ai #programming #devops

9 min read

A Sign on a Climbable Wall: Why AI Agents Need Accountability, Not Just Guardrails

#ai #agents #governance #security

5 min read

How Ethics Emerged from Episode Logs — 17 Days of Contemplative Agent Design

#ai #python #ethics #programming

7 min read

Freedom and Constraints of Autonomous Agents — Self-Modification, Trust Boundaries, and Emergent Gameplay

#ai #security #python #design

6 min read

Porting Game Dev Memory Management to AI Agent Memory Distillation

#ai #python #memory #gamedev

6 min read

Where to Put a Coding Agent's Knowledge — and How to Make It Stick

#ai #claudecode #agents #devtools

18 min read

My Agent's Memory Broke — A Day Wrestling a 9B Model

11 min read

Not Reasoning, Not Tools — What If the Essence of AI Agents Is Memory?

#discuss #ai #agents

14 min read

What Actually Happens When You Write an Article with AI — 20 Dialogues, 2 Hours, a Completely Different Core

#ai #writing #productivity

8 min read

Do Autonomous Agents Really Need an Orchestration Layer?

#discuss #ai #programming

9 min read

Every LLM App Is Just a Markdown-and-Code Sandwich

#ai #python #programming

7 min read

Natural Language as Architecture — Controlling an Autonomous Agent with Prompts, Memory, and Fail-Safe Design

#ai #python #security

8 min read

The Install and Hope Problem with MCP Tools

#claudecode #mcp #ai #devtools

5 min read

Prompt-Based Alignment Has a Ceiling — 3-Model Prisoner's Dilemma Evidence

#ai #alignment #benchmark

10 min read

click() Betrayed Me 3 Times — Playwright Magento Battle-Tested Patterns

#playwright #python #magento #automation

6 min read

I Built an AI Agent from Scratch Because Frameworks Are the Vulnerability

#ai #python #security

16 min read

I Made a Paper I Don't Understand Run in the Browser — Active Inference Claude Code

#claudecode #numpy #streamlit #python

10 min read

Obsidian's Official CLI Is Here — No More Hacking Your Vault from the Back Door

#obsidian #cli #ai #productivity

9 min read

Stop Using Default Settings — 10 Claude Code Configs That Actually Work

#ai #cli #devtools

9 min read

I Found the Shortest Route to Ship My Claude Code Skills to 57,000 Developers

#ai #opensource #skill #agentskills

5 min read

I Gave Claude Code a Simulator and It Started Tapping and Taking Screenshots on Its Own

#ai #ios #xcode #mcp

7 min read

Kimi Wrote 8,500 Lines, Blamed Me for Delegating, Then Claimed to Be Claude

#discuss #ai #kimi #solodev

7 min read

Embedding Memory into Claude Code: From Session Loss to Persistent Context

#discuss #ai #mcp #productivity

6 min read

Kimi Killed 4 of Claude's Best Ideas — An AI Peer Review in Practice

#discuss #ai #kimi

7 min read

I Asked Gemini How Its Own Deep Research Works — Half of It Was Inaccurate

#ai #gemini #chatgpt #claude

9 min read

Building a Claude Code Kimi K2.5 Hybrid Environment: Opus Designs, K2.5 Implements

#claudecode #kimi #ai #agents

6 min read

Offloading AI's Weak Spots to Shell Scripts — Designing, Building, and Publishing a Skill Audit Command

#ai #programming #testing

11 min read

The Real Power of Claude Code Isn't Code Generation — It's Autonomous Context Orchestration

#discuss #ai #developertools #productivity

6 min read

Before and After Alignment — I Typed 'Hello' Into a Base Model and Got an Anime Review

#discuss #ai #machinelearning

5 min read

Running Claude Code from iPhone via SSH + tmux

#ai #termius #ssh #tailscale

10 min read

Cursor to Zed: Disabling Built-in AI for a CLI-First Setup

#zed #ai #cursor #editor

11 min read

The Day I Told Claude Code 'You ARE the LLM'

7 min read

What I Learned from Breaking My AI Pipeline for Two Days Straight

#ai #programming

6 min read

I Tried an Opus Orchestrator and Killed It — The ROI of Multi-Agent Systems

#ai #productivity

6 min read

15 Days of Skill Sprawl in Claude Code — Lessons from 3 Audits

10 min read

Never Trust LLM Output — 6 Defenses from Building a PDF-to-Anki CLI

#ai #anki #python

10 min read

Growing a Zenn Writing Environment from Scratch with Claude Code

#claudecode #zenn #textlint #python

6 min read

A Beginner's First 10 Days of Real Development with ECC

#ai #git #beginners

9 min read

Tracking Skill Origins in Claude Code — How I Classified 79 Skills with origin Metadata

#claudecode #ai #devtools #cli

3 min read

loading...