Thang Doan

When Phone Approval of Coding Agents Earns Its Complexity

Approving an agent decision from your phone sounds like a flex. Most of the time it is the wrong answer. Here is the one scenario where it pays off.

When to Map One Agent to One Worktree to One Branch

Sharing a workspace across agents feels efficient. It is the most expensive shortcut in a multi-agent workflow. Here is when one-to-one mapping earns its overhead.

I Installed Herdr to Fix My Agents. The Second Week Was Slower.

You install Herdr and suddenly you can see all your agents in one table. That is not the same as having agents that coordinate. The two problems are easy to confuse.

How I Stopped Losing Track of Which Agent Owns Which Task

Two agents can share a branch and you survive. The third agent is where ownership breaks. Here is the mapping I use to keep one agent tied to one task, one branch, one worktree.

How I Close My Laptop Without Losing My Coding Agent Mid-Run

An agent that takes forty minutes to finish a task cannot survive a laptop sleep. The fix is not faster agents. It is a session that outlives the client.

Why Your Worktree Directory Becomes Unmanageable Past Ten Active Tasks

One worktree is a parallel workspace. Ten worktrees is a lifecycle problem. Without naming, cleanup, and ordering rules, the directory becomes a graveyard.

Why Your Coding Agent Terminal Becomes Unreadable Past Three Panes

Two agents fit on one screen. Three forces you to switch. Four means you stop reading the output. The terminal is not the right unit for parallel agents.

I Tried Running Four Agents in Parallel. Twenty Minutes of Silence Is Where It Broke.

Three agents pushed branches on schedule. The fourth stalled on a quota ceiling and I did not notice for twenty minutes. The fix was not better agents. It was inspectable ones.

The MCP You Connected Is Invisible to Your Spawned Agent

MCPs you wire into your main session do not cross into spawned sub-agents. The agent reports Linear unavailable and invents its scope from text. Here is what actually propagates and what to do about it.

Why I Stopped Running Parallel Agents in the Same Repo

Three agents in one workspace collide on branches, stash each other's state, and inherit leftover files. Sibling worktrees plus a fresh base branch fix all three. Here is the setup.

I Spawned 4 Autonomous Agents. Review Is Where It Broke

Spawning agents is the easy part. Reviewing their autonomous decisions without a logging contract is the bottleneck. Here is the format that fixed it for me.

When AI Agents Split Your Codebase Into Too Many Files

The agent's PR has six new files for one small feature. A utils file for one function. A types file for two interfaces. An index.ts that re-exports everything. None of it earns its keep.

AI Agents Are the Most Expensive Junior Devs You'll Ever Hire

The same over-engineering reflex that costs a team two days of junior time now costs you an hour of agent output, multiplied across every PR. The tax rate did not change. The volume did.

Boring Over Clever: Why Simple Code Wins at 3am

Clever code impresses in code review. Boring code survives production. Here's why the most experienced developers write the most unexciting code, and why you should too.

Stop Building What Already Exists: The Lazy Dev Mindset

Every junior dev I've met tries to build what already exists. Here's how the "laziest solution that works" mindset, backed by real code, makes you a better developer, faster.

Git Hotfix in Production: The GitHub Flow Emergency Playbook

It's Friday 5pm. Production is broken. Your feature branch is half done. Here's exactly what to do, and what NOT to do, using GitHub Flow.

Git PR Workflow & Code Review: How to Get Your PRs Reviewed Fast

A PR titled 'changes' gets ignored. Learn how to write pull requests that get reviewed fast and how to respond to feedback like a professional.

Why Your Multi-Agent Workflow Keeps Colliding

Two agents in two threads share the filesystem but not context. They make decisions based on stale state. The fix is one fresh agent per task with isolated context.

The Discipline Stack That Makes Agent Output Trustworthy

The reliability problem is not the model. It is the missing disciplines around the model. Brainstorm before code. Plan before implement. Test before pass. Verify before claim. Review before merge.

Why Your Agent Forgets Step 5 by Step 12

A ten-task plan drifts by task four. The model is not forgetful. The plan is too coarse. Tasks that look discrete to you blur together in execution. Smaller tasks with sharper edges fix it.

The One Rule That Stops Agents from Inventing Tests

An agent writes code, then writes a test for it. The test passes immediately. It proves nothing. The fix is one iron law: no production code without a failing test first.

Why Your Agent's First Fix Attempt Is Usually Wrong

An agent hits a bug, guesses a fix, reports done. Two hours later the same bug reappears because the first fix treated a symptom. Random fixes add bugs. The fix is the iron law: no fixes without root cause.

Why Your AI Agent Says "Done" When the Work Isn't

The most expensive bug in AI-assisted engineering is not the agent writing bad code. It is the agent writing code, claiming it works, and you trusting the claim without checking.

Why Your Agent Starts Coding Before You Finish Talking

You give the agent a one-sentence task. It immediately writes code. The code is plausible. It is also not what you wanted. The discipline teams skip first: brainstorm before code, no exceptions.

Why Your Agent's 'Done' Cannot Be Trusted Without Command Output

The most expensive failure mode in AI-assisted engineering is the agent reporting success without verifying. The fix is one rule: no completion claim without fresh command output.

Stop Describing Your Codebase. Show It to the Agent.

You described your project in English. The agent produced generic code that matches none of it. Stop describing. Mention the file. Run the command. Watch the output change.

Understanding AI Model Parameters: Temperature, Top-K, and Top-P

A plain-English guide to the settings that control how AI generates text, temperature, top-K, top-P, max tokens, and stop sequences, and when to tweak each one.

The System Prompt Your Coding Agent Actually Needs

A coding agent without a custom system prompt is a chatbot that happens to write code. The default prompt runs the agent like a generic assistant. Yours should run it like an engineer on your team.

Choosing the Right Claude Model: Haiku, Sonnet, and Opus Explained Simply

A beginner-friendly guide to Claude's three AI models: Haiku, Sonnet, and Opus. Learn which one to pick for your task and how to avoid wasting your usage limits.

Cutting Designer-Engineer Handoff From Two Weeks to Two Hours

The bottleneck is not engineering speed. It is the loop between intent and working interface. Claude Design compresses that loop, but only if you use it the way that actually pays off.

Adding Context and Breaking Down Complex Prompts

Learn how to give AI the right background information and how to split complex tasks into smaller, manageable steps using chaining and aggregation.

Getting Started with Claude Cowork: A Simple Setup Guide

A beginner-friendly walkthrough for setting up Claude Cowork, connect your files, tools, and preferences so Claude can do real work alongside you.

Adapting Spec-Driven Workflow to Your Team Without Burning Time

The default workflow does not fit your team. You can abandon it, or customize it. Three levels: project config (easy), custom schemas (flexible), global overrides (power). Most teams should stop at level one.

How to Describe Code Changes Without Rewriting the Spec Every Time

Specs go stale because updating them is painful. The fix is a format that lets you describe changes without rewriting. ADDED, MODIFIED, REMOVED. Three sections, no full-rewrite pain.

Gemini 3 Prompting Best Practices

A practical guide to getting the best results from Gemini 3 models, core prompting principles, Flash-specific tips, and ready-to-use templates.

Adding Spec-Driven Workflow to an Existing Project Without Burning Months

You inherited a codebase with no specs. The agent guesses at everything. The trick is not to spec backward. Spec going forward, one change at a time. Thirty-minute setup, propose-apply-archive loop.

Why Your AI Agent Ignores Your Team's Conventions

You typed the rule in chat three times and the agent still broke it. Chat is for the current task. Project rules live in AGENTS.md. Yours is empty.

Edit, Restore, or Handoff: The Three Reset Moves Every Agent User Needs

Three things break an agent session: a wrong turn, a drifted conversation, a thread that grew too big. Three different moves fix each one. Picking the wrong move wastes time.

Why Phase-Locked Workflow Breaks When the Agent Learns Mid-task

Traditional workflow forces phases. You cannot go back. You cannot test individual pieces. Real work does not happen in phases. The fix is actions you take anytime, in any order.

Prompt Iteration: Fixing Prompts That Don't Work

Your first prompt rarely gives perfect results. Learn three practical techniques to iterate and improve your prompts until the AI gives you exactly what you need.

Prompting Basics: How to Write Clear Instructions for AI

Learn the fundamentals of writing prompts that get good results, from asking questions to giving tasks, setting constraints, and controlling the format of AI responses.

Why Your AI Agent Gets Dumber After 30 Minutes

The model did not get worse. Your context filled up. Every token in the window influences every output, and the noise is now winning.

Which Artifact to Update When the Agent Gets Confused

You open a change folder. Four artifact files. You do not know which to update when the design was wrong. Most teams guess. Artifacts drift out of sync. The fix is knowing what each is for.

Why Your AI Coding Agent Forgets Every Decision You Made

You asked for the same feature twice and got two different implementations. The problem is not the model. It is that your decisions only live in chat history.

Zero-Shot vs Few-Shot: Teaching AI with Examples

Discover the difference between zero-shot and few-shot prompting, and learn how giving AI a few examples can dramatically improve the quality of its responses.