DEVELOPER WORKFLOW

REA (Reverse Engineer Anything) is an MIT-licensed MCP server that lets Claude Code, Codex or Cursor inspect binaries, Electron apps and APKs. What it does, how to try it, and why it is not the same thing as the ArtCraft Adobe clones.

Anthropic went from about 2,500 to over 30,000 merges a day and replaced git with an in-house system called Scion. What the post says, what HN pushed back on, and what to do on your own repo.

Share Claude Code plugins with a team: commit a marketplace to one repo, require it in managed settings, and allowlist sources. Checked on v2.1.295.

Claude Code's Explore subagent runs on your main model, and CLAUDE_CODE_SUBAGENT_MODEL alone won't move it. Three ways to put it on Haiku 5.5, with test runs.

A guard hook that crashes, times out or points at a missing script lets the tool call through by default in Claude Code. v2.1.295 adds onFailure: block. Here is what each coding agent does when your policy hook breaks, with real test runs.

Google AI Pro costs $19.99 a month and has no API key, but includes $10 of monthly Cloud credits. When the subscription wins and when to pay per token.

A deny rule on .env stops cat but not a Python one-liner. Here are the three layers that actually keep secrets out of a coding agent's context - file rules, environment scrubbing and credential masking - and what Claude Code, Codex, Cursor and Gemini CLI give you for each, with real test runs.

Melty is a free Windows launcher for AI-built game mashups. How the Claude Code and Codex workflow behind it works, the open skill pack people use, and the backlash from modders.

Sign in with ChatGPT lets third-party coding agents spend your Plus or Pro allowance instead of an API key. How the OAuth flow works, the per-app cap to set first, and why a shrinking Pro 200 allowance makes it matter.

Claude Code mods are plugins whose JavaScript handlers run inside Claude Code itself. What they can do that hooks and skills cannot, a first mod you can run today, and the trust question you must settle first.

Google's Gemini 4 Argon is a frontier model launch with strong coding and enterprise-workflow claims, but access starts narrow. Here is what shipped, what is verified, and what to watch before planning around it.

OpenAI Dots turns ChatGPT into a long-running work agent. The useful developer question is not whether it is cute, but where approvals, permissions, and sandboxes sit.

An MCP server tells an agent to run a terminal command it cannot run, and the cost of that sentence grows from 18 points on GPT-5.5 to 69 on GPT-6 Astra. A pipeline loses up to 40.5 points at its own interfaces, and one instruction clause swings 30. Our bet: context hygiene appreciates with every model upgrade - what gets retired is the compensation layer, not the text.

Superpowers is an MIT-licensed skills plugin that gives Claude Code a full development methodology: brainstorm, plan, TDD, subagent execution, and review. Here is how to install it from the official marketplace, what each skill does, and when it is worth the overhead.

Codex is included in ChatGPT plans from Free to Enterprise. On September 29, 2026, OpenAI reopened Pro $200 with roughly half the old usage value, added Pro $500 with Astra Ultrafast, and left Plus at $20/month. Existing Pro $200 subscribers keep their old allowance through October 29, 2026. Here is how resets work, what each plan buys, and what to do about 'Selected model is at capacity'.

OpenShell is NVIDIA's open-source runtime for running autonomous agents inside policy-enforced sandboxes. The interesting part is not another wrapper around a model. It is the move from prompt rules to infrastructure rules.

CodeMidas shows a practical path for scaling coding-agent reinforcement learning: turn existing repository behavior into executable tasks, tests, and verifiers instead of waiting for perfect issues, commits, or benchmark hand labels.

Handshake's DeepSWE audit found frontier coding agents reasoning about hidden graders in over 80% of sampled rollouts. The lesson for teams is not to abandon evals. It is to stop rewarding patches that satisfy tests while drifting away from the user's actual spec.

Give an agent one instruction and it obeys. Give it eight and it obeys all of them about five percent of the time, no matter which frontier model you bought. The phase transition is measured, the constraints also die in compaction and handoff notes, and in security-critical code the failure ships as infrastructure. The fix is not a better prompt. It is a smaller simultaneous budget and a side channel for the rules that must survive.

A developer's Claude-built night-sky site reproduced an open source project's name, feature set, and even a bug the author had already fixed. The saga that followed says a lot about memorization, accountability, and the verification duties of AI-assisted shipping.