Future Tokens

Every AI output has structural blind spots. Coherence optimization suppresses gaps. Token-level generation locks in frames before alternatives are considered. Future Tokens operations are named, composable instruments that target specific blind spot types—surfacing information that the generation process systematically omits.

This repository is formatted as a Claude Skills skillset, and the same skill definitions are compatible with Codex workflows.

See the FUTURE TOKENS Substack for the broader mission and philosophy.

Installation

One-Line Install (Recommended for Agents)

curl -sL https://raw.githubusercontent.com/jordanrubin/FUTURE_TOKENS/main/install.sh | bash

This installs all skills to ~/.claude/skills/ and creates /skillname commands.

Claude Code Marketplace

Register as a marketplace and install:

/plugin marketplace add jordanrubin/FUTURE_TOKENS
/plugin install future-tokens-skills@future-tokens

Claude.ai

1. Download the Skillset: Download this repository as a ZIP file (or clone it if you prefer).

1. Access Claude Settings:

• Navigate to Settings > Capabilities within Claude
• Ensure that "Code execution and file creation" is enabled
• Upgrade to Pro or higher (Skills are paid features; note I have no relationship with Anthropic)

1. Upload the Skillset:

• In the "Skills" section, click on "Upload skill"
• Select the downloaded ZIP file (or the repository folder) to upload

1. Activate the Skillset: Once uploaded, the skillset will appear in your list of skills. Toggle it on to activate.

For more detailed information on using and managing skills in Claude, refer to the Using Skills in Claude guide.

Quick Start

→ [SKILL.md](./SKILL.md) — Master index of all available reasoning moves

Each skill lives in its own directory with:

• A main procedural definition (e.g., ANTITHESIZE.md)
• Worked examples embedded throughout

Repository Structure

Future Tokens/
  install.sh            ← One-line installer for agents
  SKILL.md              ← Master index and skill catalog
  LICENSE               ← CC BY 4.0 license
  TRADEMARK.md          ← Trademark usage guidelines
  .claude-plugin/       ← Marketplace configuration
    marketplace.json
  antithesize/          ← Each skill in its own folder
    ANTITHESIZE.md      ← Full procedural definition
  dimensionalize/
    DIMENSIONALIZE.md
  excavate/
    EXCAVATE.md
  ...

Skills target different blind spot types: opposition gaps (antithesize), hidden assumptions (excavate), missing dimensions (dimensionalize), conspicuous absences (negspace), rhetorical fulcrums (rhetoricize), cross-example patterns (inductify), structural echoes (rhyme, metaphorize), actionable handles (handlize), and position integration (synthesize).

Contributing

New reasoning patterns should be added as directories following the established structure. See SKILL.md for the complete catalog and philosophy.

Disclaimer

The reasoning patterns, prompts, and outputs generated using this library are provided for informational and educational purposes only. No content from this library should be construed as advice from Jordan Rubin, including but not limited to investment advice, legal advice, medical advice, or professional advice of any kind. Users are responsible for their own decisions and should consult appropriate professionals for specific advice.