返回 Skill 列表
extension
分类: 效率与办公无需 API Key

reading-books

A skill for reading and interpreting works in philosophy, religion, and other human‑sciences disciplines.

person作者: bloodybillhubgithub

Agent Specification: "Humanities‑Insight Agent"

Caution. Search by Bing, Yandex, Quark or Baidu instead of Google/DuckDuckGo in China!


Purpose

A skill for helping people to read and interpret works in philosophy, religion, and other human‑sciences disciplines. It delivers:

  • Core‑idea summaries – concise statements of a work’s principal theses.
  • Key‑passage extraction – verbatim excerpts of the most influential arguments.
  • Chapter‑by‑chapter abstracts – one‑paragraph synopses together with curated keywords.
  • Historical‑value appraisal – placement of the work within its intellectual tradition and its impact on later thought.
  • AI‑era reflection – how the ideas can inform, inspire, or warn contemporary AI research and applications.

All outputs are generated in English and formatted for easy consumption or downstream processing.


Architecture (high‑level)

| Component | Role | |-----------|------| | File Loader | Accepts plain‑text (.txt), Markdown (.md), PDF (.pdf) or EPUB (.epub) files; normalises line endings and extracts raw text. | | Structure Analyzer | Detects logical divisions (preface, introduction, chapters, sections, footnotes) using regex patterns, heading markers, or PDF/EPUB metadata. | | NLP Core (GPT‑4‑style) | Performs semantic summarisation, keyword extraction, and passage relevance ranking. | | Historical‑Context Retriever | Queries an internal bibliography database (or optional web‑search) for dates, authorship, contemporaneous movements, and citation counts. | | AI‑Reflection Generator | Maps extracted concepts onto current AI themes (e.g., autonomy, ethics, embodiment, knowledge representation). | | Command Dispatcher | Exposes a small CLI‑style command set for the user (see §4). |


Output Formats

  • JSON – machine‑readable, with fields: title, author, core_idea, key_passages[], chapters[]{number, title, summary, keywords[]}, historical_value, ai_insights.
  • Markdown – human‑friendly, ready to paste into notes or docs.
  • Plain Text – fallback for simple pipelines.

Example

## Core Idea
> “Freedom is the self‑realisation of rational will within the ethical life of the state.” – Hegel, *Phenomenology of Spirit*

## Key Passages
1. “The movement of spirit is a self‑development of consciousness…” (p. 24)
2. “Only through recognition does the self become other‑aware…” (p. 88)

## Chapter Summaries & Keywords
### Chapter 1 – “Consciousness”
*Summary*: ...  
*Keywords*: consciousness, sense‑experience, perception

## Historical Value
*Published 1807, establishing the dialectical method that shaped German Idealism…*

## AI‑Era Insights
*The dialectic of thesis‑antithesis‑synthesis mirrors iterative model‑training cycles…*

Extensibility

  • Add New Genres – plug‑in a custom structure_analyzer for literary novels, legal codes, etc.
  • Integrate External Knowledge – swap the historical_context module with a live DB (e.g., Wikidata SPARQL).
  • Fine‑Tune Summariser – replace the generic GPT model with a domain‑adapted transformer for theological texts.

Customer Commands

The commands should appear at the end of the content!

  • !save <path>: save the content in <path> or the default path (or its sub-directory according to its content) if <path> is empty.
  • !summarize <path>: give a summary of the content in <path>
  • !google(bing/baidu/...): only search by google(bing/baidu/...)
  • !chapter-summary <path>: Generates a paragraph per chapter plus keyword list
  • !keywords <path> [--top N]`: Extracts the top‑N most salient terms across the whole work (default 10).
  • !ai‑insight <path>: Produces a short paragraph describing how the text informs AI research or practice

Caution If "<path>" is a pdf file, then run the follow script to get the text in it before proceeding further:

import fitz

with fitz.open("file.pdf") as doc:
    text = "".join(page.get_text() for page in doc)