LevelBench-100

levelcfo/levelbench-100

Empirical benchmark isolating base model capability vs. CLI harness architecture (Claude Code, Codex CLI, Gemini CLI, Antigravity CLI) and deterministic verification playbooks across 100 multi-file accounting close checks, 15 SMB close workbooks, and 14 browser-automation portal gates. Demonstrates that while raw frontier models top out at 71.0% with a 28.6% bad-entry rate, multi-file CLI harnesse

Language not detected No licence declared

View on GitHub → Find it in the atlas

Install

git clone https://github.com/levelcfo/levelbench-100.git

Repository reader

Explore before you clone

Read the project’s README, skills and supporting Markdown here. Rendered view is shown first; source is one click away when you need the exact text.

Fetched from GitHub
Open on GitHub
Loading the project’s README…

Public files are loaded on demand from the repository. The atlas does not copy or modify them.

Screenshot

Screenshot or social card for LevelBench-100
GitHub's social card — no screenshot was found for this project, fetched from source rather than copied into this repository.

Platform support

Derived from the project's own README, install route, CI configuration and release assets — not from running it.
PlatformRuns hereOn what evidence
WindowsNoNo evidence of support anywhere in the project.
WSL2NoNo evidence of support anywhere in the project.
macOSNoNo evidence of support anywhere in the project.
LinuxNoNo evidence of support anywhere in the project.
DockerNoNo evidence of support anywhere in the project.

Where it came from

Named by 1 of the 39 lists the atlas merges.

How it is classified

One topic, 4 harnesses. Each link is a prerendered page of everything else in that slice.

Observability & EvalsClaude CodeClaude / AnthropicCodex / OpenAIGemini / Google

Related in Observability & Evals

Closest first — projects that plug into the same harnesses, then the ones the most lists agreed on.

  • ClawMetryvivekchand/clawmetrySelf-hosted, local-first observability and kill switch for coding agents, reading Gemini CLI sessions alongside Claude Code, Codex, Cursor…
  • Codeburngetagentseal/codeburnFree, local tool that tracks AI coding token usage and cost across 31 tools and agents by model, project, and task. Worth including because…
  • onWatchonllm-dev/onWatchOpen-source Go CLI that tracks AI API quota usage across 7 providers (Anthropic, OpenAI, GitHub Copilot, MiniMax, and more). Works with Claude Code…
  • Tokscalejunhoyeo/tokscaleCLI tool for tracking token usage from AI coding agents (OpenCode, Claude Code, OpenClaw, Codex, Gemini CLI, Cursor IDE, AmpCode, Factory Droid)…
  • Clawd on Deskrullerzhou-afk/clawd-on-deskby Ruller_Lulu - A desktop pet that reacts to your Claude Code sessions in real-time — thinking, typing, juggling, sleeping, and more. Yep. It's…
  • CodexBarsteipete/CodexBarmacOS menu bar app showing usage stats for Codex, Claude, Cursor, Gemini without login. On-device parsing
  • claude-tapliaohch3/claude-tap(3,303 ⭐) - Local proxy and trace viewer for inspecting coding-agent API requests, prompts, tool calls, streaming responses, and token usage
  • ClaudeBartddworks/ClaudeBarby tddworks - A macOS menu-bar app that surfaces remaining usage quota for Claude, Codex, Gemini, Copilot, and other AI coding providers at a…