Mindframes

Structured thinking frameworks for AI agents, based on established research in cognitive psychology, philosophy, and engineering creativity.

A meta-routing skill (mindframes) classifies problems and selects the right thinking framework automatically. Four specialized sub-skills can also be invoked directly.

Skills

Skill	Frameworks	Use when
mindframes	Meta-router	You want the agent to pick the right framework
structured-evaluation	Multi-frame review + dialectical synthesis	Evaluating, comparing, deciding
creative-transformation	Transformation operators + contradiction resolution	Improving or redesigning existing artifacts
investigative-reasoning	Abductive reasoning + Socratic inquiry	Root cause analysis, debugging, migration planning
divergent-convergent	Divergent/convergent cycling + two-cycle exploration	Open-ended exploration, brainstorming

Installation

Claude Code (recommended)

Register as a plugin marketplace:

# From Claude Code
/plugin install mindframes@llupin4/mindframes

Or add the repo as an additional directory:

claude --add-dir /path/to/mindframes

Skills in skills/ are discovered automatically.

Manual install (any agent)

Copy individual skill folders into your agent's skills directory:

# Claude Code — personal (available in all projects)
cp -r skills/mindframes ~/.claude/skills/mindframes

# Claude Code — project-scoped
cp -r skills/mindframes .claude/skills/mindframes

# OpenAI Codex CLI
cp -r skills/mindframes ~/.codex/skills/mindframes

Install just the sub-skills you want, or install everything. The meta-router (mindframes) invokes the sub-skills by name, so install it alongside whichever sub-skills you plan to use.

Claude.ai / API

Upload skill folders via the Skills API. See Skills API Quickstart.

Usage

Let the router pick

"Help me think through whether we should migrate from REST to GraphQL"

The mindframes skill triggers, classifies this as an EVALUATE + WELL-DEFINED + HAS ARTIFACT problem, and routes to structured-evaluation.

Invoke directly

"/structured-evaluation Compare our three caching strategies"

"/investigative-reasoning Why did latency spike after last week's deploy?"

"/creative-transformation Redesign our notification system"

"/divergent-convergent We need a new approach to onboarding — I don't know where to start"

Structure

mindframes/
├── README.md
├── LICENSE
├── CLAUDE.md
└── skills/
    ├── mindframes/              # Meta-router
    │   └── SKILL.md
    ├── structured-evaluation/   # Standalone sub-skill
    │   └── SKILL.md
    ├── creative-transformation/ # Standalone sub-skill
    │   └── SKILL.md
    ├── investigative-reasoning/ # Standalone sub-skill
    │   └── SKILL.md
    └── divergent-convergent/    # Standalone sub-skill
        └── SKILL.md

Each sub-skill is self-contained and works independently. The meta-router invokes them by name — install it alongside whichever sub-skills you plan to use.

Background

These skills are original workflows inspired by established reasoning, design, and creativity literature, including divergent/convergent thinking, abductive inference, dialectical analysis, contradiction-resolution patterns, and question-led inquiry.

The two-cycle exploration pattern in divergent-convergent is informed in part by the Design Council's Double Diamond, available under CC BY 4.0.

See also: Cognitive Prompting in LLMs (Kramer, 2025), Sketch-of-Thought (Aytes et al., 2025), AutoTRIZ (2024).

License

Apache 2.0