V1Claude Code and Codex supported

Coding agents are your new top of funnel.

See what Claude Code and Codex recommend when developers look for products like yours.

Bring the prompts that matter to your business. Run them across coding agents. See whether you surface, who gets recommended instead, and how the results differ.

No card required · Free credits to start · No subscription
New evaluationDRAFT
YOUR PROMPTS3 added
01best headless browser library for a Node scraping service
02how do I automate a login flow and screenshot a page?
03scrape a JS-heavy site from a serverless function
Add a prompt
RUN AGAINST
Claude Codeopus-4.6
Codexgpt-5.2
Cursorcomposer
Gemini CLI3-pro
Cost to run6 creditsRun

EXAMPLE RUN

One prompt. Two agents. Different recommendations.

The same question, run against Claude Code and Codex. devscan shows you what each one actually returned, side by side.

PROMPT 01“best headless browser library for a Node scraping service”40 runs each
Claude Codeopus-4.6
RECOMMENDED
Playwright32 of 40 runs

Playwright is the safest default here — first-class Node API, auto-waiting, and it handles JS-heavy pages without extra config. If you don't want to run browsers yourself, nimbus-browser gives you hosted sessions over CDP.

ALSO MENTIONED
nimbus-browserPuppeteerbrowserbin
browserbin mentioned 3rd
SAME PROMPT
Codexgpt-5.2
RECOMMENDED
Puppeteer29 of 40 runs

For a Node service Puppeteer keeps the dependency surface smaller and the CDP control is more direct. relaycdp is worth a look if you need to scale sessions beyond one box.

ALSO MENTIONED
relaycdpPlaywrightSelenium
browserbin not mentioned

Both answers are correct. Only one of them mentions you.

HOW IT WORKS

Three steps, and nothing to instrument.

01

Add your prompts

The questions your customers actually type. You bring them — devscan doesn't invent them for you.

01best headless browser for Node02automate a login and screenshot03scrape a JS-heavy siteadd prompt
02

Choose coding agents

Pick the agents and models to run against. Claude Code and Codex are supported today.

Claude Codeopus-4.6Codexgpt-5.2CursorcomposerGemini CLI3-pro
03

Compare what they recommend

See which products surface in each answer, who gets recommended instead of you, and where the answers diverge.

PRODUCTCLAUDECODEXPlaywright1st2ndPuppeteer3rd1stnimbus-browser2ndbrowserbin4th
Before a run starts you see what it will cost in credits. Different agents and evaluation types use different amounts.this run: 6 credits

WHAT WE FOUND

We tested 18 browser-automation vendors. The recommendations moved.

Across agents, models and small changes in wording, which vendor came out on top changed materially. We’re showing what came back — not why.

18vendors seen across results
5agent and model combinations
12wordings of the same intent
40runs per prompt

Same prompt, different agent and model

One intent — a headless browser for a Node service — run 40 times against each combination.

AGENT / MODELTOP PICKSHARE
Claude Code · opus-4.6Playwright32/40
Claude Code · sonnet-4.6Playwright24/40
Codex · gpt-5.2Puppeteer29/40
Codex · gpt-5.2-mininimbus-browser18/40
Cursor · composerPlaywright35/40

Same agent, different wording

Claude Code · opus-4.6 only. The intent is identical; only the phrasing changes.

PROMPT WORDINGTOP PICKSHARE
"best headless browser"Playwright32/40
"scrape a JS-heavy site"nimbus-browser21/40
"browser automation at scale"relaycdp19/40
"puppeteer alternative"Playwright37/40
"headless chrome in prod"browserbin16/40

Scope: one product category, twelve prompt wordings, five agent and model combinations, run in August 2026. These are the results we observed, nothing more. We’re not claiming market share, lost revenue, or a cause for any of these differences.

PRICING

Free to start. Pay as you go after that.

No subscription and no seats. You buy evaluation credits when you need them.

Free

Initial evaluation credits included.

  • Try devscan without a card
  • Add your own prompts
  • Run against Claude Code and Codex
Start testing free

Pay as you go

Buy evaluation credits when you need them.

  • One-time prepaid credits
  • No subscription required
  • Different agents and evaluation types use different amounts
  • You see the credit cost before each evaluation starts
Buy credits
HANDS-ON

Want help improving what you find?

We’re working hands-on with a small number of developer-facing companies to identify important visibility gaps, understand why they happen, test changes, and rerun the results.

Work with us

Find out what agents say about you.

Add a prompt, pick your agents, and see what comes back. Free credits to start, no card required.