AI-Literacy-Superpowers Core Overview

AI-Literacy-Superpowers is a plugin ecosystem designed for Claude Code and GitHub Copilot CLI, implementing a complete AI Literacy framework workflow. Key components include constraint engineering, agent orchestration, literate programming, CUPID code review, and a three-layer execution loop. Its core idea is 'environment shapes behavior'—building a 'habitat' to guide and constrain development instead of relying on individual discipline.

Background & Core Philosophy

LLMs amplify existing engineering practices: teams with good practices benefit more, while those without may suffer. The project is inspired by Christopher Alexander's '无名之质' (design for users) and Richard Gabriel's 'habitability' (code as a 'living' place). It aims to create a habitat that shapes development behavior.

Three-Layer Execution Loop

Based on Birgitta Boeckeler's Harness Engineering:

Advisory Loop

Real-time non-blocking feedback: constraint gate (HARNESS.md violations), markdownlint, drift detection, snapshot staleness check, reflection prompts, secret scanning, GC checks.

Strict Loop

Merge-blocking enforcement: CI workflows (PR constraints, weekly GC, mutation tests) and agent pipeline (orchestrator, approval gates, TDD agent, code reviewer, integration agent).

Investigative Loop

Regular cleanup: weekly GC rules, composite learning (REFLECTION_LOG.md curation), constraint audit, health checks.

Key Components

Skill System

18 modular skills grouped into:

• Code quality: Literate Programming (Knuth's rules), CUPID code review (Terhorst-North's attributes).
• Constraint engineering: Harness Engineering, Context Engineering, Constraint Design, Garbage Collection.
• Security: GitHub Actions supply chain, dependency audits, Docker Scout, secret detection.
• Observability: Constraint observability, convention extraction, cross-repo orchestration.

Agent Team

Coordinated agents: Orchestrator (pipeline coordinator), Spec-writer (update specs), TDD agent (failing tests), Code-reviewer (CUPID/literate perspective), Integration agent (PR/CI/merge), Constraint discoverer/executor/GC/auditor, Evaluator (AI literacy assessment).

Practical Tools & Observability

Command System

Slash commands: /superpowers-init (setup), /superpowers-status (dashboard), /harness-constrain (add constraints), /reflect (reflections), /assess (AI literacy evaluation).

Four-Layer Observability

• Operational rhythm: health snapshots, staleness prompts.
• Trend visibility: snapshot differences, multi-cycle views.
• Telemetry export: OTLP-compatible metrics.
• Meta-observability: self-checks (snapshot timeliness, learning flow, GC effect).

Model Routing

MODEL_ROUTING.md maps agents to model tiers (strongest/standard/fast) for cost control.

Academic & Practical Foundations

Cognitive Science

• Andy Clark: Predictive processing & embodied mind.
• Edwin Hutchins: Distributed cognition.
• Lucy Suchman: Plans vs situated actions.
• James Gibson: Affordances.
• John Boyd: OODA loop.
• Donella Meadows: System thinking.

Practical Sources

• Christopher Alexander: Pattern Language.
• Richard Gabriel: Software habitability.
• Donald Knuth: Literate programming.
• Daniel Terhorst-North: CUPID.
• Birgitta Boeckeler: Harness Engineering.
• Addy Osmani: Code Agent Orchestra.
• 2025 DORA Report: AI as practice amplifier.

Application Scenarios & Value

Suitable For

• Teams wanting systematic AI workflows.
• Organizations needing cross-project consistency.
• Teams explicitizing implicit conventions.
• Developers pursuing human-AI collaboration.

Value

Turns AI Literacy theory into executable practice. Plugin design allows gradual adoption without full workflow overhaul.

Summary & Outlook

AI-Literacy-Superpowers shifts from AI code generation to workflow orchestration and environment constraints. It enables efficient human-AI collaboration.

As AI grows, such frameworks will be critical. This project provides a systematic reference for teams adopting AI-assisted development.