Skip to content
Back to the index

MazeBench

mazebench.comInfrastructure

PulseGate's liveness check found it on 14 Sep 2026; it is registered on GitHub and PyPI and has been in the index since 7 Jul 2026. How this is checked

MazeBench is an open-source tool for benchmarking coding agents through interactive ice-maze puzzles. It includes a web-based puzzle site, a world editor for creating custom mazes, and a CLI for running benchmarks. It is designed for AI researchers and developers working on agent evaluation.

Inferred · not functionally tested

Open SourceMITWebCLI
MazeBench preview
Visit mazebench.com

Overview

5 features

Purpose: Benchmarking and evaluating coding agents using interactive maze puzzles and custom environments.

Inferred · not functionally tested

Audience: ai researchers

Inferred · not functionally tested

Functions: analytics

Inferred · not functionally tested

Interfaces: API: unknown · MCP: unknown · CLI: indicated (inferred, not tested) · Self-hosting: unknown

Recorded constraints: pricing: open_source · license: MIT · platforms: CLI, WEB · deployment: browser, cli

Constraint provenance is unknown; confirm requirements with the publisher.

Record sources: mazebench.com · github.com. These links do not verify the individual claims.

In the Agent evaluation & testing space, MazeBench takes a focused approach. Inferred · not functionally tested: It focuses on benchmarking and evaluating coding agents using interactive maze puzzles and custom environments. Inferred · not functionally tested: It is built as an open-source project for ai researchers. Basis unknown · not verified: The project is open source (MIT). Basis unknown · not verified: It runs on the web and the command line.

MazeBench first shipped in 2026. Development happens publicly on GitHub with 348 commits in the last 90 days. Inferred · not functionally tested: Among its 5 catalogued features are puzzle site, world editor, and benchmark runner.

Summary written by a language model from the project’s public pages.

Tasks: Inferred · not functionally tested

  • Puzzle site
  • World editor
  • Benchmark runner
  • Coding agent evaluation
  • Local web server

Topics: Inferred · not functionally tested

Tags
agent-benchmarkmaze-puzzlesworld-editor

JSON profile · Text profile · Access guide

Built with & integrations

Framework
Next.js
Hosting
Cloudflare
Runs on
BrowserCLI

Trust & compliance

License
MIT
Public signals
HTTPSOpen SourceFree tierGitHubActive maintenance

Indexing history

What PulseGate has recorded for this listing

Nothing recorded for this listing in this window.

Frequently asked questions about MazeBench

What does MazeBench do?
Inferred · not functionally tested: MazeBench focuses on benchmarking and evaluating coding agents using interactive maze puzzles and custom environments. It is catalogued under Agent evaluation & testing on PulseGate.
Who should use MazeBench?
Inferred · not functionally tested: MazeBench is an open-source project built for ai researchers.
Does MazeBench have a free plan?
Basis unknown · not verified: Yes — MazeBench is open source under the MIT license and free to use.
What platforms does MazeBench run on?
Basis unknown · not verified: MazeBench runs on the web and the command line.
Is MazeBench still maintained?
PulseGate's liveness check found it on 14 Sep 2026. Its GitHub repository shows 348 commits in the last 90 days.
When did MazeBench launch?
MazeBench first shipped in 2026.
Is MazeBench open source?
Basis unknown · not verified: Yes — MazeBench is open source under the MIT license, developed on GitHub.

Similar projects

Closest matches by what these projects do