Study which tools coding agents pick for you

A team ran about 17k coding-agent sessions to see which databases and services Claude Code, Codex and Cursor choose.

Difficulty
AdvancedOur rating
Time to build
Not reportedOur estimate
Monthly cost
Not reported
Revenue
Not reportedWe only show revenue the source states
Engagement
300 HN points153 comments · fetched Oct 15, 2026

What they did

They ran agents on real repos with different personas and logged the services they installed.

How it went

Claude Code rarely searched the web, Codex almost always did, and the agents often disagreed.

Worth knowing

The authors sell growth services to dev-tool companies, and say so.

Builder recipe

Our own starting point, written from the source. Adapt it to your setup.

Stack

  • Coding agents
  • Sandboxed repos

Steps

  1. Define the question and success criteria.
  2. Let agents search, run or summarize in parallel.
  3. Verify key results yourself.
  4. Write up findings with sources.

Starter prompt

Tags

Read the original

Related use cases

The Daily 5.

Five real AI agent use cases each morning, each with a recipe and a source. Plus a Sunday roundup. Free.