The LGTM (Page 92)

The LGTM

Home
Agentic Coding
Claude Code
Codex

Sign in Subscribe

DeepSeek V4 Is the First Open Model Release This Year That Actually Treats Context Length Like an Engineering Constraint

DeepSeek V4 Is the First Open Model Release This Year That Actually Treats Context Length Like an Engineering Constraint

Open models keep advertising giant context windows the way cloud vendors advertise unlimited storage: technically true, operationally slippery. The hard part is not printing 1M tokens on a model card. The hard part is building a model and serving stack that do not melt once an agent spends an afternoon

24 Apr 2026 5 min read

OpenClaw’s Cron Tool Leak Turns Agent Orchestration into FinOps Whether Maintainers Like It or Not

The most mature thing happening in agent infrastructure right now is not better prose generation. It is cost paranoia. That is a healthy sign. Once teams move agents from novelty to recurring operations, every hidden token, every inflated tool schema, and every unnecessary background call stops being an implementation detail

24 Apr 2026 4 min read

OpenClaw’s Discord Thread Fix Is a Good Reminder That Multi-Agent Systems Fail at the Handoff, Not the Demo

The most revealing bugs in agent software are rarely the ones that crash. They are the ones that succeed halfway. The child session starts, the logs look green, the worker probably did the job, and then the answer never shows up where the human is actually waiting. That is not

24 Apr 2026 4 min read

OpenClaw’s Latest Beta Makes a Bigger Bet on Delegation, but It Also Shows Where the Control Plane Is Getting Stricter

OpenClaw’s Latest Beta Makes a Bigger Bet on Delegation, but It Also Shows Where the Control Plane Is Getting Stricter

The most interesting OpenClaw releases lately have stopped trying to impress you with one shiny feature. They read more like platform negotiations. Maintainers loosen one capability where users clearly need it, then tighten three other boundaries so the system does not become a permission-shaped mess. That is exactly the pattern

24 Apr 2026 4 min read

OpenClaw’s Pi/Codex Runtime RFC Says Agent Platforms Are Finally Treating Policy Drift Like a Real Systems Problem

OpenClaw’s Pi/Codex Runtime RFC Says Agent Platforms Are Finally Treating Policy Drift Like a Real Systems Problem

The easiest way for an agent platform to lose credibility is not to crash. It is to behave differently depending on which runtime happened to answer the call. Same user prompt, same surface, same platform logo, subtly different tool behavior, auth forwarding, fallback handling, or delivery semantics. That is how

24 Apr 2026 4 min read

GPT-5.4 Looks Like the Point Where Vision Pipelines Get to Delete a Lot of Defensive Code

GPT-5.4 Looks Like the Point Where Vision Pipelines Get to Delete a Lot of Defensive Code

Most model upgrades add capabilities. The interesting ones delete code. That is the real takeaway from Microsoft’s new benchmark on GPT-5.4’s ability to return pixel-level coordinates from engineering drawings. On the surface, this is a niche vision post about bounding boxes in electrical single-line diagrams. In practice,

24 Apr 2026 5 min read

xAI's Remote MCP Push Says Grok Wants to Plug Into Your Agent Stack, Not Replace It

xAI's Remote MCP Push Says Grok Wants to Plug Into Your Agent Stack, Not Replace It

xAI keeps telling the market that Grok is a model. Its docs keep revealing something more useful: Grok is being turned into a connector. That distinction matters. In 2026, the AI vendors that win serious developer usage are not the ones with the loudest benchmark screenshots. They are the ones

24 Apr 2026 4 min read

Mastra 1.27.0 Is Chasing the Agent-Operator Layer, Not Just the Framework Layer

Mastra 1.27.0 Is Chasing the Agent-Operator Layer, Not Just the Framework Layer

Mastra’s latest release is interesting for a reason that has very little to do with version numerology and a lot to do with where the agent-framework market is headed. @mastra/[email protected] is not selling one grand abstraction. It is tightening the operator layer, the part of

24 Apr 2026 5 min read

Safer Treats Agent Shell Access Like a Capability Problem, Which Is Exactly the Right Instinct Right Now

Safer Treats Agent Shell Access Like a Capability Problem, Which Is Exactly the Right Instinct Right Now

AI coding tools are entering their permissions era, and not a moment too soon. For the last year, the category has behaved as if terminal access were just another convenience feature. Let the agent edit files, install packages, restart containers, talk to Kubernetes, touch databases, maybe push a branch, and

24 Apr 2026 4 min read

Claude Code’s Next Trick Is Delegating to a Pool of Codex Workers Instead of Doing the Whole Job Itself

Claude Code’s Next Trick Is Delegating to a Pool of Codex Workers Instead of Doing the Whole Job Itself

Multi-agent coding has spent the last year sounding like a research demo genre. One agent plans, another implements, a third reviews, and somewhere in the middle a human tries to remember which branch is the real one. Most of that work has been high on spectacle and low on operational

24 Apr 2026 4 min read

Doby Is a Small, Sharp Rebuttal to the Idea That AI Coding Needs More Context Instead of Better Navigation

Doby Is a Small, Sharp Rebuttal to the Idea That AI Coding Needs More Context Instead of Better Navigation

There is a lazy answer to most AI-coding quality problems right now: give the model more context. Bigger windows, more retrieval, more MCP tools, more documents, more screenshots, more everything. Sometimes that works. Sometimes it just creates a larger room for the model to get lost in before it edits

24 Apr 2026 4 min read

Copilot’s Pull Request Chat Update Is Really About Giving the Agent More Review Surface Area

Copilot’s Pull Request Chat Update Is Really About Giving the Agent More Review Surface Area

GitHub’s latest Copilot pull-request chat update looks small enough to ignore. That would be a mistake. On the surface, the April 23 changelog entry is about better pull-request context, improved summaries, and more review-oriented chat behavior when Copilot is given a PR as context. But the deeper story is

24 Apr 2026 4 min read

← Newer Posts Page 92 of 115 Older Posts →

The LGTM © 2026

Sign up

Powered by Ghost