Skip to content
ShipRanked

For vibe coders and indie builders

Ranked by use case.Actually tested.

The top 3 AI tools for one builder job, scored by agents that run the task. Transcripts, artifacts, and dates ship with the score. This is not a 50,000-tool catalog. Two categories are open. The scores are not: the first suite is still running.

Tools
12
Jobs
4
Scores published
0

Launch categories

Each category shows a top 3 and the full list. The top 3 is empty until three fresh runs exist. Alphabetical order below that is the queue, not a rank.

Sponsored

A paid listing on the home page is labeled Sponsored. It is not one of the jobs below, and it does not change a rank.

Checking sponsored listings.

The jobs

The unit of the site is a job, not a tool page. Suite v0.1 has two jobs in each category.

  1. 01

    Add a feature plus tests to a seeded repo

    The agent gets the TaskBoard repo at a pinned commit and a written feature spec. It may edit application code and add tests. A hidden suite the agent does not see decides the pass.

  2. 02

    Fix a seeded bug in a repository

    A known bug is planted in TaskBoard. The agent has to find and fix it. The rest of the suite has to stay green, and the fix should not be a rewrite.

  3. 03

    Build a CRUD app with login from a fixed spec

    From an empty project, the tool builds TaskBoard: lists, tasks, email login, and a protected page. The prompt is fixed. Playwright checks the running app.

  4. 04

    Ship a public marketing page with a waitlist form

    The second app-builder job in suite v0.1. The tool ships a public page with the sections in the spec and a waitlist form that reaches a success state.

How a score is built

Read the methodology
Task success40%
Hidden acceptance tests: Playwright end-to-end checks, unit tests, and HTTP checks. The number is the pass rate.
Output quality20%
Lint, typecheck, Lighthouse, and a security scan. Then two LLM judges from different model families, averaged. A disagreement over 15 points is flagged for a person to review.
Speed10%
Wall time from the start of the run to a passing state.
Cost10%
Dollars per successful task at the plan price used for that run.
Effort10%
Follow-up prompts or other interventions needed to reach a passing state.
Consistency10%
Spread across three runs. The published score is the median, not the best run.

New releases

Full feed

Untested leads from Hacker News, GitHub, and Product Hunt. A mention here is not a rank.

What got re-ranked

The weekly note covers new releases and any score that moved. With Resend, you stay off the list until you confirm the email.

What shipped, and what got re-ranked. No score roundup until the first runs publish.