your ad here · $499/moyour ad here · $499/moyour ad here · $499/moyour ad here · $499/moyour ad here · $499/moyour ad here · $499/moyour ad here · $499/moyour ad here · $499/moyour ad here · $499/moyour ad here · $499/moyour ad here · $499/moyour ad here · $499/mo
Launching October 6, 2026
kodwai

kodwai

Get scored on how well you use your AI coding agent

Visit website

FreeAIAI Agents8 impressionslaunches Oct 6

Anyone can one-shot green tests now. The interesting part is how well you use the agent.

kodwai gives you real engineering challenges that you solve on your own machine with your own agent: Claude Code, Cursor or Codex. No browser sandbox.

kodwai challenge bookshelf-rest-api pulls the problem, starter code and tests into a fresh git repo and starts the timer. kodwai submit sends it off.

You get a 0 to 100 score across three axes: Direction (how you steer, verify and decompose), Outcome (what actually shipped) and Lift (how far you beat a solo AI). Every signal cites evidence from your transcript, commits and test runs.

Free for developers, with a public leaderboard. 3 scored submissions on us, then bring your own Anthropic key.

AIAI Agents
  • Solve real coding challenges on your own machine with your own AI agent
  • Scores how well you direct the agent, not memorization
  • Fully free – bring your own Claude Code, Cursor or Codex
  • Public leaderboard to rank your AI‑assisted coding
  • Used by developers at Google, Meta, Apple, Microsoft, Amazon, Stripe, Netflix

Kodwai lets developers solve coding challenges with their own AI coding agents and get scored on how they direct the agent.

for
Developers who use AI coding assistants like Claude Code, Cursor or Codex.
pricing
freemium
license
open source
kodwai/cli 0 0TypeScriptupdated 6 hours ago
works withClaude CodeCursorCodex

Key features

6 features of kodwai
  • Agent-agnostic CLI — Run challenges locally with Claude Code, Cursor, Codex or any terminal-based agent.
  • Real-world challenges — Each task includes starter code, tests and hidden traps to mimic production work.
  • Multi-axis scoring — Scores out of 100 across Direction, Outcome/Rubric and Lift with evidence from transcripts and commits.
  • Public leaderboard & weekly leagues — Compete in weekly resets and see how you rank against other developers.
  • Zero-sandbox execution — All work happens on your machine; only challenge files are uploaded for scoring.
  • Evidence-backed feedback — Score breakdown shows moments, symbols and citations so you can review and improve.

Use cases

  • Demonstrate proficiency in steering AI coding agents for hiring assessments
  • Practice handling hidden edge-cases and traps in backend services
  • Benchmark personal AI-assistant workflow against peers in weekly leagues

kodwai vs alternatives

all alternatives →
kodwaiFinePieces CopilotDify.AIluoluoAI
Best forScoring AI-assisted coding workflowsAI-agent coding assistanceIn-IDE AI coding assistanceLLMOps and app buildingModel endpoint management
PricingFreemiumFreeFreeFreeOne time fee
DevHunt upvotes175633052
LaunchedOct 2026Jan 2023Mar 2024Jan 2023Sep 2026
  • kodwai vs Fine: Focuses on building software with AI agents rather than scoring agent-driven workflows
  • kodwai vs Pieces Copilot: Provides an AI coding sidekick integrated into the IDE, not a challenge-scoring platform
  • kodwai vs Dify.AI: Open-source LLMOps platform for building AI-native apps, not a coding-challenge scoring service
  • kodwai vs luoluoAI: Aggregates model endpoints and pricing, unrelated to agent-driven coding challenges

kodwai FAQ

How do I start a challenge?+

Run `npx @kodwai/cli@latest challenge <slug>` which signs you in, downloads the problem, starter files and tests, creates a git repo and starts the timer.

What agents are supported?+

Kodwai works with Claude Code, Cursor, Codex and any other terminal-based agent, though those three are officially tested.

Is there a cost to use Kodwai?+

The platform is free for developers; after three scored submissions you need to provide your own Anthropic API key.

What does the score represent?+

A total out of 100 split into Direction (how you steer), Outcome/Rubric (what shipped) and Lift (how you beat a solo AI), each backed by evidence.

How is my code submitted?+

After solving, run `npx @kodwai/cli@latest submit` which uploads your code, git history, test runs and agent transcript from the challenge folder.

Can I see why I got a particular score?+

Yes, the results list up to six annotated moments with symbols (!!, ?, ??) and citations from your session.

Summarized by DevHunt from kodwai.com · Sep 29, 2026. Details may change; check the official site.

Trending launches