Discover the Best AI Resources

nohurry/Opus-4.6-Reasoning-3000x-filtered

Filtered subset of the OPUS 4.6 parallel corpus that isolates reasoning-related translation examples and removes 979 refusals, providing a cleaner 3,000×-filtered dataset for training or evaluating NLP models focused on reasoning in translation.

Crownelius/Opus-4.6-Reasoning-3300x

Cleaned reasoning dataset of problem→thinking→solution triplets derived from Opus 4.6, provided in Parquet with ~2,160 cleaned rows (original 3,305). Filters remove empty/short/refusal/non‑substantive responses; hosted on Hugging Face under Apache‑2.0.

JobOps

Find and rank jobs across major boards, auto-tailor a resume per role using an LLM-driven pipeline (RxResume v4), and track application emails — all as a self-hosted Docker pipeline for privacy-conscious users.

Claude Code Game Studios

Turns a single Claude Code session into a coordinated game-development studio by providing 49 specialized AI agents, 72 slash-command skills, automated hooks and path-scoped rules. Includes tiered roles, engine-specific agent sets (Godot/Unity/Unreal) and templates to keep design, QA and release in sync.

oMLX

Local LLM inference server for Apple Silicon that exposes an OpenAI-compatible API and a macOS menubar app. Uses continuous batching and a two-tier KV cache (RAM + SSD in safetensors) to persist context across restarts, enabling practical multi-model serving and fast local coding workflows.

OmniRoute

Acts as an OpenAI‑compatible local and cloud gateway that routes requests across 100+ LLM providers with smart routing, load balancing, retries and fallbacks. Adds policies, rate limits, semantic caching and observability for reliable, cost‑aware inference in Docker, Electron or npm installs.

pg_durable

Runs durable, checkpointed SQL workflows inside PostgreSQL so long-running data and AI pipelines can resume after crashes without external orchestrators. Provides a SQL DSL, in-process background worker, and Postgres-backed state—useful for embeddings, ETL, scheduling, and fan-out jobs when you can install extensions.

React Doctor

Scans a React codebase and produces a 0–100 health score plus actionable diagnostics across state & effects, performance, architecture, security, accessibility, and dead code. Auto-adapts to framework and React version, supports Next.js/Vite/React Native, a CLI, GitHub Action, and agent integrations to teach coding agents.

WebWorldData

Provides 1.06M web interaction trajectories (state, action, next_state) represented primarily as A11y trees for training browser world models and web agents. Covers diverse real‑web domains, English/Chinese pages, and long contexts (up to 30K tokens); residual PII and dynamic content may limit reproducibility.

Tolaria

Desktop app for managing markdown-based knowledge bases with a files-first, git-first workflow. Works offline, uses plain markdown + YAML frontmatter for portability, and includes AI-agent integrations and agent configuration to organize context, memory, and procedures for assistants.

Agent Skills

Encodes production-grade engineering workflows (spec, plan, build, test, review, ship) as reusable "skills" so AI coding agents follow consistent development practices. Packaged as per-skill SKILL.md files and slash commands for integration with agents and CLIs. Suited for teams embedding engineering guardrails into agent-driven dev workflows.

forge

Provides a reliability layer for self-hosted LLM tool-calling and multi-step agent workflows. Adds guardrails — rescue parsing, response validation, retry nudges, and a synthetic respond tool — and ships a Drop-in OpenAI-compatible proxy plus a WorkflowRunner for structured loops.