Skip to content

ai-literacy-superpowers

A plugin marketplace for Claude Code and GitHub Copilot CLI shipping opinionated tools for the AI Literacy framework — harness engineering, agent orchestration, decision archaeology, governance, and model evaluation.

Get Started · Browse Plugins · Quick Install

Plugins in this marketplace

Plugin What it does Get started
ai-literacy-superpowers The flagship. Harness engineering, agent orchestration, literate programming, CUPID code review, compound learning, and the three enforcement loops. 30 skills, 13 agents, 24 commands. Tutorial
model-cards Researches and authors Mitchell-extended model cards from a model name. Tiered source strategy with refusal-on-unconfirmed-existence honesty rule. Tutorial

The marketplace is on track to ship more sister plugins as the framework grows. New plugins land under /plugins/ with their own tutorials, how-to guides, reference, and explanation pages.

Want to contribute to the plugins themselves? See Contributing — the conventions, harness constraints, TDAD test discipline, common pitfalls, and first-PR checklist that govern day-to-day development.

For the methodology behind the plugin's test discipline, see TDAD — Test-Driven Agentic Discipline applied to skills, agents, and commands, with the four-layer cost-cadence architecture that scales the discipline across the project.

Add the marketplace

The marketplace ships multiple plugins. Add the marketplace once, then install whichever plugins you need.

Claude Code:

claude plugin marketplace add Habitat-Thinking/ai-literacy-superpowers

GitHub Copilot CLI:

copilot plugin marketplace add Habitat-Thinking/ai-literacy-superpowers

Install a plugin

After adding the marketplace, install one or both plugins:

# Claude Code
claude plugin install ai-literacy-superpowers     # the flagship
claude plugin install model-cards                  # the sister

# Copilot CLI
copilot plugin install ai-literacy-superpowers@ai-literacy-superpowers
copilot plugin install model-cards@ai-literacy-superpowers

Each plugin's commands, agents, and skills become available immediately in any project session.

What each plugin ships

ai-literacy-superpowers — the flagship

The flagship plugin implements harness engineering: surrounding AI code generation with deterministic tooling, agent-based review, and periodic entropy checks so AI-assisted development stays trustworthy at scale.

It gives you:

  • 30 skills — domain knowledge for security auditing, constraint design, context engineering, fitness functions, model sovereignty, decision archaeology, and more
  • 13 agents — autonomous workers for orchestration, enforcement, garbage collection, code review, governance auditing, adversarial spec/code review, decision archaeology, and TDD
  • 24 commands — slash commands for harness lifecycle, assessment, portfolio assessment, reflection, governance, onboarding, affordance inventory, decision-archaeology mapping, and convention management
  • 3 enforcement loops — advisory at edit time, strict at merge time, investigative on schedule

In a fresh project, run /superpowers-init for the full habitat or /harness-init for just the constraint machinery.

Browse the ai-literacy-superpowers docs →

model-cards — Mitchell-extended model card research

Aimed at evaluators researching new models. Produces 10-section cards (Mitchell et al.'s 9 canonical sections plus an "Operational Details" section for consumer-evaluator audiences), with per-claim citations and tiered source provenance.

It gives you:

  • /model-card create <name> — researches a single model with a human-in-the-loop review checkpoint before the file is written
  • /model-card seed — bulk-populates the library with cards for 14 frontier LLMs from major providers
  • A read-only research agent with no Write access — the trust boundary that prevents hallucinated claims from landing on disk unreviewed
  • An existence-check refusal rule — the agent refuses to produce a card when it cannot confirm the model exists via tier-1 (provider docs) or tier-2 (HuggingFace), preventing authoritative-looking cards for hallucinated models

Browse the model-cards docs →

Documentation structure

Each plugin in the marketplace has its own landing page with tutorials, how-to guides, reference material, and explanation pages organised using the Diataxis framework.

Plugin Documentation
ai-literacy-superpowers Landing · Tutorials · How-to · Reference · Explanation
model-cards Landing · Tutorial · How-to · Reference · Explanation

See the Plugins index for the canonical list.

Old URLs

The previous flat layout (/tutorials/, /how-to/, /reference/, /explanation/) was reorganised on 2026-04-29 to support multiple plugins in the same marketplace. Old URLs redirect to the new plugin-scoped paths automatically via jekyll-redirect-from.

Why a marketplace, not a monolith

Every plugin in this marketplace addresses a distinct discipline that benefits from AI Literacy practice — harness engineering, model evaluation, future plugins covering further domains. Shipping each as its own plugin lets teams adopt incrementally:

  • A team focused on AI-assisted code quality can install ai-literacy-superpowers and skip model-cards until they need to catalogue models.
  • A team building an evaluation pipeline can install model-cards without needing the full harness machinery.
  • Future plugins for adjacent disciplines (cost governance, prompt evaluation, agent observability) can land in the same marketplace without inflating the flagship plugin.

Shared infrastructure — the marketplace listing, the docs site, the CI conventions, the trust-boundary patterns — lives at the repository level. The plugins themselves stay focused.