AG

atsushi-green/ds-ai-coding-skills

Data science & ML

This page was translated automatically. Technical details may differ from the original. View the original

91 stars Quality 65 Trend 29MIT

A collection of AGENTS.md configurations, skills, and prompts tailored for Data Scientists to maximize productivity with AI coding assistants.

Overview

This is a repository template that integrates with GitHub Copilot / Copilot Agent Mode / Copilot Cloud Agent and Claude Code for secure and consistent data analysis work. This repository supports both and . While having a common docs/agent/ documentation and skill system, it provides dedicated instruction files for each agent. This design avoids large, all-in instruction files and allows tokens to efficiently reference only the information they need. In Claude Code, it is provided as .claude/skills//SKILL.md (which can also be launched in slashes like /plan-analysis), while Copilot continues to provide it as .github/prompts/.prompt.md as task-execution skills. Claude Code and GitHub Copilot skills correspond to two different types. - : .claude/skills/ /SKILL.md ⇔ .github/skills/ /SKILL.md。 Contents are the same except for the relative path notation in skill links. - (Disable-model-invocation: true): .claude/skills/ /SKILL.md ⇔ .github/prompts/ .prompt.md. Converts frontmatter format and synchronizes while maintaining and converting Copilot's ${input:...} placeholders and CLAUDE.md⇔AGENTS.md cross-references. - uv run python scripts/sync_agent_docs.py --check — Determines only the presence or absence of diffs without writing (run_quality_checks.sh / already built into CI. Only differences for normal skills are terminated code 1). - uv run python scripts/sync_agent_docs.py --from claude (or --from github) — **Reflects on the other side. Direction is required (does not automatically check by mtime).

Overview (machine translated)


Links Contents
📊 Diagnostic Chart Gallery Each *-diagnostics skill has specified that it must be output, and the actual data is output and arranged in sequence
📝 AGENTS.md and Skills for Data Scientists Router documentation (AGENTS.md) and design of basic skills such as Python, SQL, data processing, and visualization
📝 Continued: Skills for Data Scientists: Analysis Methods × Diagnostic Visualization Set *-diagnostics Design and Operation Example of a Skill Group (a mechanism that displays figures and values together)

Setup

Prerequisites

  • Python 3.11
  • The UV must be installed

Installation

Main Commands

Directory Structure

Designing the Agent Instruction System

This repository supports three platforms: GitHub Copilot, Claude Code, and Codex. To ensure the same rules apply regardless of the tool, we have a policy of placing the substance of instructions in only one place.

Router documentation (one content provided in two files)

Each tool has a set file name to read, so although there are two files, the content is identical. ## Hard Rules and later matches to verify with CI verification by scripts/sync_agent_docs.py --check.

File Reading Tools Roles
CLAUDE.md Claude Code Hard rules, common commands, conventions, skill routing. Only the ## Skills at the end of skill import) is unique to Claude
AGENTS.md GitHub Copilot (VS Code / CLI / coding agent) and Codex Same content as above. Only the introduction at the beginning is unique

Skills and Auxiliary Files

Path Reading Tools Role
.claude/skills/*/SKILL.md Common to all three tools Detailed procedures by task (Python, SQL, data processing, visualization, diagnostics, etc.). No mirror
.agents/skills Codex Symbolic link to .claude/skills
`.

This page was translated automatically. Technical details may differ from the original.

Recommended Tools

Try a different keyword or remove a filter.

Install

npx skillfish add atsushi-green/ds-ai-coding-skills
New to MCP? Start with the mcpskill.org guide →