Cangjie Skill: Distill Books into Executable AI Skills

Key Takeaways#

• Cangjie Skill turns high-value books into executable AI skill packs rather than simple summaries.
• Its RIA-TV++ method combines whole-book analysis, parallel extraction, triple verification, structured skill construction, Zettelkasten linking, pressure testing, and delivery into agent skill directories.
• The repo includes templates, methodology docs, extractors, generated skill pack examples, and multilingual documentation.
• It is best treated as a resource and methodology for agent-skill authors, not as a hosted tool.

What it is#

Cangjie Skill is a methodology repository for distilling books and long-form materials into reusable AI skills. The project starts from a simple problem: people read valuable books but often fail to apply the ideas in real work. A flat summary compresses content, but it does not tell an agent when to use a concept, how to apply it, what its boundaries are, or how to test whether the skill was triggered correctly. Cangjie Skill aims to convert useful methodology from source material into independently callable skill packs that an AI agent can reuse during tasks.

The RIA-TV++ pipeline#

The README describes a seven-stage process. First, whole-book comprehension uses structural, interpretive, critical, and applicability analysis to build an overview. Second, specialized extractors pull frameworks, principles, cases, counter-examples, and glossary items in parallel. Third, triple verification checks each candidate for supporting passages, predictive power against new questions, and non-obviousness. Fourth, verified material becomes RIA++ structure: original quote, interpretation, book cases, future trigger scenarios, executable steps, and boundaries. Fifth, Zettelkasten linking records dependencies and contrasts between skills. Sixth, pressure tests check whether the skills respond correctly to prompts and bait questions. Seventh, the output is delivered as digest files, indexes, glossary entries, skill files, and test prompts.

When to use it#

Use Cangjie Skill when you want to create agent-ready knowledge from books, founder letters, strategy texts, product handbooks, or other high-value source material. It is especially useful when the goal is action: decision support, writing guidance, strategic analysis, coaching prompts, or domain-specific operating practices.

Why builders should care#

Agent skill ecosystems need better source discipline. Cangjie’s verification and pressure-testing stages are useful patterns because they stop every appealing passage from becoming a skill. The method asks whether a unit has evidence, a trigger scenario, execution steps, and known boundaries. That makes the output easier for an agent to invoke and easier for a human to audit.

Limits#

Cangjie Skill is not a one-click book ingestion product. It is a structured resource for building skill packs, with templates and examples. Users still need source rights, careful extraction, review, and a compatible agent environment such as Claude Code or Cursor skills.