How Stage 0 Adler Analysis Extracts Book Structure and Critique in Cangjie‑Skill

Stage 0 Adler analysis extracts a book's structure and critique through four systematic steps—Structure, Interpretation, Critique, and Applicability—that output a structured BOOK_OVERVIEW.md artifact ready for downstream skill extraction.

The kangarooking/cangjie-skill pipeline transforms raw book content into reusable knowledge using Mortimer Adler's "analytical reading" methodology. This entry stage dissects long‑form material into a rigorous overview that powers subsequent skill extraction stages, ensuring every downstream component inherits both structural clarity and critical boundaries.

The Four Steps of Stage 0 Adler Analysis

According to methodology/01-stage0-adler.md, Stage 0 implements four discrete analytical steps. Each step maps directly to a section in the final BOOK_OVERVIEW.md output.

Step 1: Structure — Identify the Book's Skeleton

The Structure step isolates the book's architectural backbone. Adler analysis at this stage demands answers to【01-stage0-adler.md†L11-L18】:

• Type of work: methodology, biography, history, etc.
• One‑sentence purpose: the author's core objective
• 3–7 primary arguments: with explicit relationships (parallel, progressive, contrast, refutation)
• Core problem: what the author seeks to solve

This step produces the foundational schema that all subsequent analysis references.

Step 2: Interpretation — Restate the Author's Reasoning

The Interpretation step captures the book's intellectual content in the reader's own terms【01-stage0-adler.md†L20-L25】:

• Recurring key terms: with author‑specific definitions
• 5–15 core propositions: restated without quotation
• Logical chain: how arguments connect and what evidence supports them

This restatement verifies genuine comprehension before any evaluation occurs.

Step 3: Critique — Surface Blind Spots Before Agreement

The Critique step enforces Adler's discipline of suspended judgment: understand fully, critique fairly, then assess【01-stage0-adler.md†L26-L35】. Required outputs include:

• Historical constraints: period‑specific limitations affecting the work
• Positional biases: career, culture, institutional affiliations shaping the author's perspective
• Unproved assumptions: premises accepted without demonstration
• Strongest possible counter‑argument: the most rigorous opposition the author faces

This step generates the "Boundary (B)" field in downstream skills, preventing unchecked propagation of author viewpoints.

Step 4: Applicability — Flag Skill‑Ready Content

The Applicability step extends Adler's method for the cangjie-skill pipeline【01-stage0-adler.md†L38-L44】:

• Candidate frameworks: checklists, principles, or procedures suitable for skillification
• Unsuitable content: pure narrative, anecdote, or context‑dependent material
• Skill count estimate: rough quantity and priority ranking for downstream stages

This bridge step transitions from analytical reading to actionable knowledge extraction.

How Stage 0 Adler Analysis Works in Practice

The extraction pipeline follows a four‑phase execution model:

1. Input ingestion: Raw book transcription (or video/podcast transcript) enters the pipeline
2. Prompt‑driven parsing: LLM calls or rule‑based parsers answer the step‑specific questions, populating placeholders like {{BOOK_TITLE}}, {{论点 1}}, {{author's definition}}
3. Template rendering: Filled values substitute into templates/BOOK_OVERVIEW.md.template, producing books/<slug>/BOOK_OVERVIEW.md
4. Quality gate validation: Output must satisfy checklist items including "主旨能用一句话说清", "骨架列出 3–7 个一级论点", and "批判阶段列出 ≥3 条作者局限"

Only documents passing this gate advance to Stage 1 and beyond.

Running Stage 0 Programmatically

The repository exposes this functionality through the cangjie-skill CLI. Below is a simplified Python wrapper demonstrating invocation:

import subprocess
import json
from pathlib import Path

def run_adler_stage(book_slug: str, source_file: Path) -> Path:
    """
    Execute Stage 0 (Adler analysis) for a given book.
    Returns the path to the generated BOOK_OVERVIEW.md.
    """
    # The `cangjie-skill` CLI expects the slug and source file.

    cmd = [
        "cangjie-skill", "stage0",
        "--slug", book_slug,
        "--source", str(source_file)
    ]
    subprocess.run(cmd, check=True)

    overview = Path(f"books/{book_slug}/BOOK_OVERVIEW.md")
    return overview

# Example usage

if __name__ == "__main__":
    overview_path = run_adler_stage(
        book_slug="deep-work",
        source_file=Path("raw/transcripts/deep_work.txt")
    )
    print(f"Generated overview: {overview_path.read_text(encoding='utf-8')}")

Execution triggers the full four‑step Adler analysis, template population, and quality validation described above.

Core Files Supporting Stage 0 Implementation

File	Purpose
methodology/01-stage0-adler.md	Defines the four analytical steps and quality gate criteria
templates/BOOK_OVERVIEW.md.template	Markdown scaffold receiving extracted data and rendering final output
SKILL.md	Pipeline overview positioning Stage 0 as the comprehension entry point
README.md	User‑facing introduction to Adler analysis concepts

These files collectively implement the architectural design that enables rigorous book structure extraction and critical review generation.

Summary

• Stage 0 Adler analysis applies Mortimer Adler's four‑step analytical reading to raw book content
• Structure, Interpretation, Critique, and Applicability steps map directly to BOOK_OVERVIEW.md sections
• Quality gate validation ensures comprehensive analysis before downstream processing
• Critique step outputs generate "Boundary (B)" fields protecting against unexamined author viewpoints
• methodology/01-stage0-adler.md and templates/BOOK_OVERVIEW.md.template encode the methodology and output format

Frequently Asked Questions

What makes Stage 0 "Adler analysis" different from standard summarization?

Standard summarization condenses content for quick consumption. Stage 0 Adler analysis enforces systematic comprehension through restatement before evaluation, surfaces author biases and limitations, and produces structured artifacts with explicit quality criteria—not merely shorter versions of input text.

Why does the Critique step require listing the author's biases before agreeing with anything?

Adler's method demands "understanding before evaluation" to prevent confirmation bias and shallow reading. By forcing enumeration of historical constraints, positional biases, unproved assumptions, and strongest counter‑arguments, the pipeline ensures downstream skills carry appropriate epistemic boundaries rather than unchecked authority.

How does the Applicability step differ from standard Adler analytical reading?

Adler's original method concluded with critique. Applicability is a cangjie-skill extension specific to knowledge extraction pipelines—it identifies which content can become reusable skills versus material that remains context‑bound, enabling efficient prioritization for subsequent automated processing stages.

What happens if a book fails the Stage 0 quality gate?

Books failing checklist validation (e.g., unclear one‑sentence purpose, fewer than 3 primary arguments, or missing critical limitations) do not advance to downstream stages. The pipeline returns the overview for revision or flags the source material as unsuitable for skill extraction, preserving output quality across the entire system.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}