mshumer
Featured
Claude-of-Duty
A browser-based first-person shooter built entirely with procedural generation and orchestrated AI agents, featuring 55k lines of Three.js/WebGL2 code and no art assets.
AI repository intelligence
FR-AI / ANALYSISWhy it is worth attention
It introduces a structured collaboration model (intent owner, evidence worker, independent verifier) and deterministic checking for OS agent memory reuse and stale-fact rejection, filling a gap in reproducible evaluation of OS agents.
Who it is for
- OS agent researchers and developers
- Academic researchers studying tool-use and memory benchmarks
- Open-source communities building Linux-based automation agents
- Engineers designing agent evaluation frameworks
Use cases
- Evaluating an OS agent's ability to diagnose system state changes across package, service, or runtime configurations
- Testing agent memory reuse policies with scoped fingerprint validation and stale-fact rejection
- Benchmarking multi-agent collaboration patterns (e.g., owner-worker-verifier) under deterministic checkers
- Reproducible integration testing of agent runners against a frozen public-development task suite
Strengths
- Domain-level tracks (system and runtime state) that are distribution-agnostic and use synthetic frozen fixtures
- Clear role-based collaboration model with observable responsibilities and deterministic checker contracts
- Public-development answer key and reference validator for integration testing without dependencies
- Methodological alignment with established benchmarks while contributing original OS-agent task content
Considerations
- Public preview – not an official OS certification suite or leaderboard; results require exact version tracking
- CC BY-NC 4.0 license restricts commercial use without separate written permission
- Requires Node.js 18+ and no held-out golden answers in the public repository (evaluator must provide private bundle)
README quick start
Quick start
Description
Research benchmark for evidence-grounded OS-agent collaboration, continuous state diagnosis, scoped memory reuse, and stale-state rejection.
Related repositories
Similar projects matched by category, topics, and programming language.
gnipbao
Featured
story-to-handdrawn-video
A Remotion-based tool that converts Chinese story text or ordered hand-drawn images into vertical hand-drawn diary-comic animation with handwritten captions, left-to-right reveals, optional page-curl transitions, and silent H.264 output for post-production dubbing.
0xwilliamortiz
Featured
ponytail-improved
Ponytail is a plugin for AI coding agents that enforces a disciplined ladder of reuse before writing code, reducing code volume by roughly 54% while preserving safety.