Stories

Better ways to work with coding agents. The tools we explore, the changes we make, and what we learn along the way.

Distinct workspace tools surround a shared, unfinished ivory-and-brass tower on a stone plinth.

What are the top ADEs for 2026

Seven agentic development environments (ADEs) for developers who run several coding agents at once, in our editorial order, compared on how they mix agents, keep their work apart, check it, and what they cost.

AgentGrid October 29, 2026

Read the story

A pencil sketch of a small house sits beside a matching ivory-and-brass model on a warmly lit workbench.

Five AI coding tools for founders and non-developers

For people who know their field better than they know code: five tools in our editorial order, compared on who owns the code, who runs the app and its data, how the work gets checked and what it costs.

AgentGrid October 29, 2026

Read the story

A mechanical gripper holds a blue component above a brass inspection fixture, beside four blue parts, some chipped.

Claude Haiku 5.5: When Cheaper Coding Agents Actually Save Money

Haiku's token bill is only the starting point. Compare accepted work after routing, review, retries and escalation.

AgentGrid October 8, 2026

Read the story

A hand fits a glowing tile into a modular ceramic sphere, an illustration of adding one focused extension.

Claude Code Mods: What They Are and How to Get Started

Make one recurring part of your coding session easier to see or control, then decide whether the customization earns its place.

AgentGrid October 6, 2026

Read the story

A small evidence packet moves from a building station to a station with an inspection lens.

Stop Being the Clipboard Between Your Coding Agents

A useful agent team needs clear ownership and a checkable handoff. Start with those before adding more workers.

AgentGrid October 6, 2026

Read the story

An illuminated checkpoint card anchors a branching trail of translucent pages.

Why Claude Code Forgets Your Plan—and How to Resume Reliably

Keep a short record of the work, then make the resumed agent check it against the project before continuing.

AgentGrid October 6, 2026

Read the story

A team's usage popover shows the Personal Claude account at 100% with a switch-to-Business hint. The footer Run as picker reassigns the Worker Build pane from Personal to Business, making it ready to pick up its next turn on that account — the Codex team lead and the Worker Research pane, which never stopped, keep running throughout.

Features

Multiple accounts per harness

Add a personal account alongside a work seat, or a handful of Codex plans, and pick which one each pane runs as. Shipped in AgentGrid v2.9.4.

AgentGrid September 30, 2026

Read the story

An illustrated workbench connects a short plan, a modular build and a separate inspection station.

Claude Code Plan Mode: When to Plan, When to Build, and When to Add a Reviewer

Use uncertainty and consequence to decide when to plan, build directly, or add a reviewer. Includes proposed prompts and an evidence handoff.

AgentGrid September 26, 2026

Read the story

Proposed AgentGrid review loop: a lead coordinates a builder and separate reviewer, with fixes and checks before your final inspection.

How to Get the Most Out of Opus 5.5: Setup, Real Use Cases, and AgentGrid

Select the right runtime and effort, give Claude a task it can verify, and keep implementation and independent review visible in one project.

September 25, 2026

Read the story

Standard API rates per million tokens: Opus 5.5 versus Fable 5.1 is $4 versus $10 for input, $20 versus $50 for output, and $0.20 versus $0.25 for cache reads.

Claude Opus 5.5 vs. Fable 5.1: What Changed, and Which Should You Use?

Opus 5.5 makes a strong case for a trial as your everyday coding model. Keep Fable 5.1 where better results on your own work justify its higher API rates.

September 23, 2026

Read the story

AgentGrid workspace with a Jev client review, passing local tests and the TypeSafe API reference.

What Is Jev and How to Get Started

Build a small issue-triage client in AgentGrid, understand its decisions, and test the review rules before making a live API call.

September 21, 2026

Read the story

AgentGrid terminals showing the pinned Laya configuration beside saved local smoke-test results, false positives and human-review routing.

Jev vs. Laya: Which AI Decision Model Should You Choose?

Choose between a hosted decision API and a local model you operate, then test the choice on the work you actually need done.

September 21, 2026

Read the story

A worker pane on Claude Code reaches a usage limit. The footer picker switches it to Codex with GPT-6-Astra, the next message is sent, a Harness switched event with a Handoff context disclosure appears, and Codex replies.

Features

Switch harnesses mid-session without losing your place

Move an open pane from Claude Code to Codex with a handoff of the conversation instead of a new pane. Shipped in AgentGrid v2.8.33.

AgentGrid September 18, 2026

Read the story

Two starting points for evaluation: Muse for ongoing personal coordination, and Grok Bot for recurring jobs with distinct owners. Both lead to testing the fit with a real task.

Muse vs. Grok Bot: Which Fits the Work You Want to Delegate?

The practical differences in organization, automation, access and control—and a way to test which one earns its place.

September 17, 2026

Read the story

Editorial illustration of tangled blue and ivory threads arranged into clear paths on a workbench

Workflows

Claude Code simplifier: clean up the code, then check the result

The Code Simplifier plugin, the /simplify command, and a practical way to review a cleanup without losing the behavior you need.

AgentGrid September 16, 2026

Read the story