Your brand hereYour brand hereYour brand hereYour brand hereYour brand hereYour brand here
Agents in the Wild

Claude writes, a second model attacks: firmware output about 2x, fewer bugs

Half their effort goes to automated adversarial review. Claude plans and writes the code, Sol tears into it with only the coding standards to go on, and it costs about $300 per developer a month.

Done withClaudeSpotted· by HN user K0balt· Oct 15, 2026

Result Fewer bugs and clearer documentation

“We use a Claude for planning and generation, sol for adversarial review. Our metrics say we are about 2x.”
From the source: Hacker News · 2026-08-03

Spotted: found in the wild. Details are as reported by the source above and have not been independently verified.

What they did

Runs a strict harness: specify everything in advance, plan, review the plan against the spec, implement and test, then a series of reviews by a different model with no context except the coding standards (test coverage, guidelines, smells, architecture). Roughly 10 percent of the effort is code generation, 40 percent documentation, planning and adversarial review, and 50 percent automated adversarial code review. Claude plans and generates; Sol does the adversarial review.

How it went

Their metrics say they’re about 2x as productive, at about $300 per developer per month, while shipping fewer bugs and clearer documentation. They credit the constant review by a different model, plus careful hand-offs between sessions.

Try it yourself with Claude
You're an adversarial code reviewer with no context except our coding standards: [paste standards]. Review this change: [diff]. Check test coverage, standards compliance, code smells, architecture and module boundaries, and anything that could break elsewhere. List each problem with its severity and a concrete fix. Don't praise anything.

A starting prompt written by Agents in the Wild. Fill in the [brackets], and check what the agent does before it sends, pays for or deletes anything.

Keep browsing

All 76 Claude use casesAll 90 Engineering use cases
Your brand hereYour brand hereYour brand hereYour brand hereYour brand hereYour brand here

Advertise on Agents in the Wild

Reach the people building and buying AI agents.

Every pageSidebars + phone strips
High-intentBuilders, not browsers
20/20Spots left

How it works

Your brand appears in rotating sponsor slots on desktop sidebars and in the scrolling sponsor strips at the top and bottom of the screen on phones, across every Agents in the Wild page. Sponsors rotate every 10 seconds so every advertiser gets fair visibility.

Pricing

Monthly rate: $250/month

Reserve your spot ($250/month) →

Billed monthly through Stripe. Cancel anytime. See the sponsor terms.

Submit an agent in the wild

Have an agent doing real work, or spotted one? Send the link and tell us the result. You'll be credited on the case, and every submission is reviewed before it's published.

By submitting, you agree to our Terms and Privacy Policy.