Scibly Skills

Open-source Agent Skills for instructional design and L&D (Learning & Development) work, built and used by Scibly. Each skill packages a proven instructional-design methodology — Bloom's Taxonomy, the ABCD model, didactic reduction, scenario design, and more — into step-by-step instructions an AI agent can follow.

What's an Agent Skill?

An Agent Skill is a folder containing a SKILL.md file with YAML frontmatter (name, description) and Markdown instructions. Coding agents and assistants (Claude Code, Codex, Cursor, OpenCode, and others) can discover and load these skills to gain specialized, repeatable workflows — without you having to re-explain the methodology every time. See the Agent Skills spec for details.

Skill catalog

Skill	What it does
needs-analysis	Synthesize SME transcripts, job descriptions, and performance data into a structured skill-gap analysis — figure out what to teach before building anything.
learning-objectives	Turn vague topics or objectives into precise, observable, testable learning objectives using Bloom's Taxonomy and the ABCD model.
didactic-reduction	Strip a topic down to its learning-essential core using the three filters from German educational science (Didaktische Reduktion): exemplary, representative, transferable.
scenario-generation	Generate realistic workplace scenarios, branching cases, and role-play briefs for practice and assessment.
assessment-design	Design quiz questions, knowledge checks, and assessments that go beyond recall, using Bloom's Taxonomy as a prompt lever for application-level items.
microlearning-design	Design the structure of short, focused microlearning sessions (3–10 min) around one behavior change.
content-editing	Review and improve AI-generated L&D content with a 5-fault-line editing framework — the "First-Draft Machine" approach.
storyboard	Turn a course outline or scene plan into a prioritized multimedia production plan — decide per scene whether it needs an infographic, animation, video, interactive scenario, or plain text, with the didactic reasoning.
instructional-prompt-engineering	Write better AI prompts for any L&D content using a 4-ingredient formula: Role, Audience, Bloom's level, Constraints, Task.

How the skills fit together

needs-analysis ──▶ learning-objectives ──▶ didactic-reduction
                                                  │
                  ┌───────────────────┬──────────┴──────────┐
                  ▼                   ▼                      ▼
        scenario-generation   assessment-design     microlearning-design
                  │                   │                      │
                  └───────────────────┴──────────┬───────────┘
                                                   ▼
                                              storyboard
                                                   │
                                                   ▼
                                           content-editing

instructional-prompt-engineering — used across every step above

You don't need to use them in order or all together — each skill works standalone, but they're designed to hand off to each other naturally (e.g. learning-objectives points to assessment-design and microlearning-design once objectives are defined).

Installation

Recommended: npx skills

Use npx skills to install skills into any supported agent (Claude Code, Codex, OpenCode, Cursor, and more):

# Install all skills into every agent detected on your machine
npx skills add scibly-dev/skills --all

# Install specific skills into specific agents
npx skills add scibly-dev/skills \
  --skill needs-analysis --skill learning-objectives \
  -a claude-code -a codex

Claude Code (manual)

Copy or symlink a skill folder into your project's .claude/skills/ directory, or into ~/.claude/skills/ for a global install:

cp -r skills/learning-objectives ~/.claude/skills/

Codex / OpenCode / Cursor (manual)

Copy the skill folder into your project's .agents/skills/ directory, or the tool's global skills directory (e.g. ~/.codex/skills/, ~/.config/opencode/skills/, ~/.cursor/skills/).

ChatGPT

ChatGPT doesn't have a native skill-folder mechanism. To use a skill there, open the relevant SKILL.md, copy everything below the frontmatter, and paste it into a Custom Instructions field or a Project's knowledge files. Include the skill's description too, so the assistant knows when to apply it.

Other / generic agent harnesses

Any harness that scans a repository's skills/ directory or reads AGENTS.md can use this repo directly — see AGENTS.md.

Repository structure

skills/
├── needs-analysis/SKILL.md
├── learning-objectives/SKILL.md
├── didactic-reduction/SKILL.md
├── scenario-generation/SKILL.md
├── assessment-design/SKILL.md
├── microlearning-design/SKILL.md
├── content-editing/SKILL.md
├── storyboard/SKILL.md
└── instructional-prompt-engineering/SKILL.md

Each skill is a single self-contained SKILL.md file in its own folder — one canonical copy per skill, installable into any supported agent.

Contributing

• One skill per folder under skills/<name>/, containing a SKILL.md file.
• name: in the frontmatter must match the folder name exactly, kebab-case, no redundant prefixes (e.g. no ai- prefix — every skill here is an AI agent skill by definition).
• description: should be a rich, specific description of when to use the skill, including concrete trigger phrases — this is what the agent uses to decide when to load it.
• All skills and documentation are written in English.

License

MIT