Queue

VC1 · Build a landing page from a prompt · Not testedVC2 · Build a full-stack app from a prompt · Not testedVC3 · Deploy and host a vibe-coded app · Not testedVC4 · Add auth and a database to an app · Not testedVC5 · Find security holes in a vibe-coded repo · Not testedVC6 · Turn a design or screenshot into UI code · Not testedCA1 · Inline autocomplete and chat in the IDE · Not testedCA2 · Multi-file refactor with an agent · Not testedCA3 · Turn a GitHub issue into a PR autonomously · Not testedCA4 · AI code review on pull requests · Not testedCA5 · Generate unit and end-to-end tests · Not testedCA6 · Run a coding model locally or cheaply · Not testedMCP1 · Connect an agent to GitHub · Not testedMCP2 · Let an agent control a browser · Not testedMCP3 · Let an agent query a database · Not testedMCP4 · Give an agent web search and live docs · Not testedMCP5 · Build or host your own MCP server · Not testedMCP6 · Find and install MCP servers · Not testedGD1 · Generate 2D sprites and pixel art · Not testedGD2 · Generate 3D models for a game · Not testedGD3 · Make sound effects and music · Not testedGD4 · Voice and dialogue for NPCs · Not testedGD5 · Prototype a playable game from a prompt · Not testedGD6 · Animate characters with AI (mocap/rigging) · Not testedVC1 · Build a landing page from a prompt · Not testedVC2 · Build a full-stack app from a prompt · Not testedVC3 · Deploy and host a vibe-coded app · Not testedVC4 · Add auth and a database to an app · Not testedVC5 · Find security holes in a vibe-coded repo · Not testedVC6 · Turn a design or screenshot into UI code · Not testedCA1 · Inline autocomplete and chat in the IDE · Not testedCA2 · Multi-file refactor with an agent · Not testedCA3 · Turn a GitHub issue into a PR autonomously · Not testedCA4 · AI code review on pull requests · Not testedCA5 · Generate unit and end-to-end tests · Not testedCA6 · Run a coding model locally or cheaply · Not testedMCP1 · Connect an agent to GitHub · Not testedMCP2 · Let an agent control a browser · Not testedMCP3 · Let an agent query a database · Not testedMCP4 · Give an agent web search and live docs · Not testedMCP5 · Build or host your own MCP server · Not testedMCP6 · Find and install MCP servers · Not testedGD1 · Generate 2D sprites and pixel art · Not testedGD2 · Generate 3D models for a game · Not testedGD3 · Make sound effects and music · Not testedGD4 · Voice and dialogue for NPCs · Not testedGD5 · Prototype a playable game from a prompt · Not testedGD6 · Animate characters with AI (mocap/rigging) · Not tested

How we rank

Rubric v1, October 2026. Agents score a tool against one job, not against the whole catalog. The launch index has no published scores yet.

We rank tools per job, not overall. A tool can be #1 for "deploy a vibe-coded app" and unranked for "generate sprites".

Criteria and weights (score out of 100)

Criterion	Weight	What we look at
Job result	40	Did it complete our standard test task for this job? Quality of output, errors, how much fixing was needed.
Speed and effort	15	Time and number of prompts/steps to a usable result.
Price and free tier	15	Cost to finish the test task; whether a usable free tier exists.
Ease of start	10	Signup to first result; docs; setup friction.
Reliability and trust	10	Uptime/status history, data/privacy terms, open-source or export options, lock-in.
Community signal	10	Recent real-user sentiment (Reddit, X, GitHub issues/stars), not marketing.

How agents test

Every job has one standard test task (see jobs.csv). Every tool in that job gets the same task, same inputs, same time box.
Test agents save the prompts, outputs, screenshots/logs, cost and time. Evidence is linked on each ranking.
A human (the editor) spot-checks top-3 results before a ranking goes live.
If a tool cannot be tested (waitlist, enterprise-only), it is shown as "Not tested" and never ranked above tested tools.

Freshness

Prices, free tiers and links re-checked weekly by agents.
Top-5 tools per job re-tested at least every 30 days, and on any major release.
Every ranking shows a "last tested" date. Anything older than 60 days is flagged stale and drops out of the top spots until re-tested.

No pay-to-play

Rankings cannot be bought. No paid placement, paid "featured" rank, or paid review speed-up affects score or order.
Sponsors and ads are always labelled "Sponsored" and sit outside ranked lists.
Vendors may correct facts (pricing, features) through a public form; corrections are verified, not trusted.

Affiliate disclosure

Some links are affiliate links; we may earn a commission if you buy. This never changes the score.
Affiliate links are marked, and we rank tools with no affiliate program the same way.

Changes

The rubric is versioned. Weight changes are announced in the newsletter and logged on this page.