Back to directory

AI Dataset Generator Skill (Claude / Codex / Antigravity / Cursor / Windsurf / Copilot)

A dataset-generation skill for AI coding assistants that turns topics, URLs, or raw files into supervised fine-tuning and preference-optimization training datasets, with verification, deduplication, and export pipelines.

AI Dataset Generator Skill (Claude / Codex / Antigravity / Cursor / Windsurf / Copilot)

AI Dataset Generator Skill packages 14 specialized sub-skills, 19 pipeline scripts, and 15 shared utility modules that generate SFT and DPO training data from a topic, a URL, or raw source files. It installs its own native format for each supported tool, including a skill directory for Claude Code and Codex CLI, an MDC rule for Cursor, a rule file for Windsurf, repository instructions for GitHub Copilot, and project rules for Cline.

Its pipeline scripts cover the full workflow from collecting and augmenting source material through verification, deduplication, coverage checking, and quality reporting, and it exports to a canonical internal schema as well as preset formats including OpenAI's chat message format, a Hugging Face dataset format, and CSV columns.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.