chradams

> In your workflow, who implements the review feedback - the review subagent or the code-writing-subagent?

I think currently my orchestrator - which spins up subagents for impl, research, review, fixes, etc.

I've been working with a single orchestrator per "project" or "feature" or whatever sort of "thing I'm paying attention to" is (I guess...chat?) and that orchestrator is told to use subagents to do any/all of this type of stuff, it's sole purpose is to just communicate with me, instantiate these types of 'work', and tell me about results or ask for input.

I am just now considering changing this though and considering something even heavier, but for the purpose of "cleaning up / polishing" lazy work by a single pass of "build thing X", this sort of orchestrator with an AGENTS.md or skill that says "always do a review pass after you finish a task by spinning up N sequential (or parallel) follow up subagents to review and another M parallel or sequential fix subagents to 'fix' those things should make a meaningful difference, if a person sees the same types of traps that their model falls into repeatedly.

You don't even need to have some complex harness. Just take a stock frontier model and its default cli or harness, ask it to edit its agents file to tell itself to always work in this manner, and you're done.

Use hooks to run a bunch of review steps after -all- code writing steps your agent does and give it the exact review criteria you just described (via git hooks, or your agent harness of choice's own hooks eg https://code.claude.com/docs/en/hooks or AGENTS.md). So after every step where the agent writes the shitty untyped string map ball of mud, your orchestrator/main thread agent that spawned the code-writing-subagent spawns a follow-up review agent automatically that is given that output, your prompt that explains what well written Go code looks like, and even the sample/golden-path code of your choice to use as a style guide.

Each time you encounter a shitty thing you hate, add a new 'review type' / 'thing to watch out for' and just ask your agent to add it to your hooks for you. This works well with Claude at least.

I have about a dozen or so hooks that run on every integration branch my agents write that review for all sorts of things from correctness to spec, performance improvement opportunities, modularity, analysis of any dependencies added, 'definition of done', UI/UX, etc.

I recently told Claude it should run the whole suite of reviews twice. I will probably go on and proceed to having it run like 5 times eventually idfk.

But the more you start asking your agents to modify their own behavior, using the native solutions offered by Cursor, or Claude, or Codex, the sooner you'll start to feel better about the results.

It's absolutely wild that your post is flagged (downvoted). Thanks for writing the article and for replying here.

Corporate taxes can tax corporate revenue or profits. Stock buybacks and dividends can be regulated or outlawed. If we don't want to force corporations to pay with more equality across ranks, we can take larger portions of revenue or profit and then fund social safety nets and welfare that are targeted at those that corporations choose not to give raises to.

There are plenty of systems answers, power just doesn't care, and neither do voter bases.

how are you liking shelley?

haven't really tried shelley much yet, I've been interacting with exe.dev over SSH and then claude cli with my sub once I'm ssh'd in. Next step I've been procrastinating is tailscale on my exe VM and mosh on the tailnet address.

Why do you think such an approach would work, but something equivalent to try to prevent mass shootings / school shooter type gun violence wouldn't? Or would you say that radical open cooperation and mutual self restraint led by governments could stop the American problem of mass shootings like the Las Vegas shooter?

I don't see how you stop terrorism (of the school shooter type, not the Islamic/political type) with inter-government cooperation. Mirror life is just another vector for school shooters but with PHDs.

Almost all for-profit large internet platform providers you use have anti-bot, anti-scraping, anti-spam, anti-abuse defenses. This is a new thing that is the same, it's anti-distillation, but its a subset of the same space.

If you have a problem with this, you should have a problem with Google, Apple, Microsoft, Amazon, Netflix, etc.

If that's also the case, then no problem.

But what Anthropic is doing here is nothing new.

Just google multi-cloud. Yes. It's a thing.