GH·RADAR

Offline water chemistry calculations with explicit units, worked formulas, CLI, Python, MCP, and browser tools

OSINT recon: check a username, email, phone or domain across 80+ public sources. CLI, API and a terminal UI.

Offline exam plans, deterministic FSRS reviews, original practice material, timed mocks, CLI, MCP, and mobile review queues

16 local LLMs on one RTX 4090 vs 6 cloud settings, 8 coding tasks designed by Claude Opus 5.5, 3 runs each: charts, videos, data

A fully local coding agent for the GPU you already own. Hardware-aware llama.cpp profiles + Qwen Code. No API key, no cloud.

Gemini CLI extension: let Gemini write and edit sheet music live in the MYV score editor

基于 Python + FastAPI 的本地 LLM API 测试工具：多站点并行测试、模型检测、思考参数、自定义 Prompt、HTML/SVG 预览与中英切换。

Offline travel and health journaling with descriptive comparisons, streaming Apple Health imports, CLI, MCP, and queued PWA check-ins

Shorten URLs without an account or API key. GET and POST API, custom aliases, optional expiration, destination lookup and deletion. JavaScript, Python and Go SDKs.

Batch organizer that moves nested image/video files up to their parent folder, with Tkinter GUI and CLI

PyAutoGUI screen image find-and-click automation helper

Self-hosted clipboard sync for macOS. A native menu bar app sends the text, formatting and images you copy to your own Cloudflare Worker + D1, and a web app lets you search and copy them from any device. Multi-user, private by default, no third-party servers.

Offline idea bank, weighted X drafts, thread splitting, bilingual variants, publishing calendar, manual analytics, CLI, and MCP

Test harness for LLM prompts and models: YAML suites, multi-provider runs, regression diffs

Unofficial Codex adaptation of pstack workflows, skills, and agents, with native Codex and Claude Code execution.

Deterministic AI benchmark for SVG spatial reasoning: can a model draw a pelican riding a bicycle?

Skateboard: Do models understand the physics of a kickflip?

🎨 Ultra-lightweight standalone theme system & 0ms tactile palette switcher for Google Antigravity 2.0

Deterministic AI agent benchmark for pizza refunds, policy compliance, escalation, tool use, CLI, API, and MCP.

Tool Errors: Do agents survive real API failures without duplicating side effects?

AI agent benchmark for cost per verified back-office task with deterministic grading.

MCP safety benchmark for poisoned tools, shadowing, least privilege, and canary exfiltration.

Coding-agent benchmark that detects fixes made by weakening tests.

Checkout Consent: Does a shopping agent spend only with clear consent?