All skills
jwynia avatar

/ebook-analysis

@7eca2c9
by J Wyniajwynia/agent-skills160 stars
20

Parse ebooks, extract concepts and entities with citation traceability, classify by type/layer, and synthesize across book collections.

Use this Skill: https://skilld.dev/gh/jwynia/agent-skills/ebook-analysis

This session only. Nothing lands on disk.

referencesrelationship-types.md

≈1.9k tokens on demand. Your agent reads this file only when SKILL.md points to it.

Relationship Types

This reference defines the relationship types used to link concepts. Use these definitions to guide relationship creation.

Overview

Relationship Meaning Directionality
INFLUENCES A affects B Directed (A→B)
SUPPORTS A provides evidence for B Directed (A→B)
CONTRADICTS A conflicts with B Bidirectional
COMPOSED_OF A contains B as component Directed (A→B)
DERIVES_FROM A is derived from B Directed (A→B)

INFLUENCES

Definition: Concept A has a causal or correlational effect on Concept B. A change in A would affect B.

When to Use:

  • A causes B to happen
  • A enables or facilitates B
  • A inhibits or prevents B
  • A is a precondition for B

Strength Values:

  • Positive (+0.1 to +1.0): A increases/enables B
  • Negative (-0.1 to -1.0): A decreases/inhibits B
  • Strong (±0.7 to ±1.0): Direct causal relationship
  • Moderate (±0.4 to ±0.6): Significant but indirect
  • Weak (±0.1 to ±0.3): Correlational or minor

Examples:

  • "Trust" INFLUENCES→ "Collaboration" (strength: +0.8)
  • "Fear" INFLUENCES→ "Risk-taking" (strength: -0.7)
  • "Shared identity" INFLUENCES→ "Community cohesion" (strength: +0.9)

Not INFLUENCES:

  • A simply mentions B (use context, not relationship)
  • A is a component of B (use COMPOSED_OF)
  • A is evidence for B (use SUPPORTS)

SUPPORTS

Definition: Concept A provides evidence, validation, or backing for Concept B. A makes B more credible or justified.

When to Use:

  • A is empirical evidence for B
  • A is an example that demonstrates B
  • A is a theoretical justification for B
  • A is a citation or reference for B

Strength Values:

  • Strong: Direct empirical evidence or proof
  • Moderate: Good example or theoretical support
  • Weak: Tangential or partial support

Examples:

  • "Research study X" SUPPORTS→ "Spaced repetition improves retention"
  • "Case study of Company Y" SUPPORTS→ "Community-led growth works"
  • "Historical precedent Z" SUPPORTS→ "Revolutions follow pattern P"

Not SUPPORTS:

  • A causes B to happen (use INFLUENCES)
  • A is part of B (use COMPOSED_OF)
  • B is derived from A (use DERIVES_FROM)

CONTRADICTS

Definition: Concept A conflicts with, opposes, or is incompatible with Concept B. They cannot both be true in the same context.

When to Use:

  • A and B make opposite claims
  • A and B recommend incompatible approaches
  • Evidence for A undermines B
  • A and B are mutually exclusive

Strength Values:

  • Strong: Direct logical contradiction
  • Moderate: Significant tension or incompatibility
  • Weak: Partial tension or contextual conflict

Types of Contradiction:

  1. Theoretical: Different claims about how things work
  2. Practical: Different recommendations for action
  3. Contextual: True in different contexts but not compatible

Examples:

  • "Move fast and break things" CONTRADICTS↔ "Measure twice, cut once"
  • "Communities need strong leadership" CONTRADICTS↔ "Communities should be self-organizing"
  • "Research shows X" CONTRADICTS↔ "Research shows not-X"

Bidirectionality: CONTRADICTS is bidirectional - if A contradicts B, then B contradicts A.

Not CONTRADICTS:

  • A is simply different from B (difference ≠ contradiction)
  • A and B apply in different contexts (may just be contextual)
  • A and B are different levels of abstraction

COMPOSED_OF

Definition: Concept A contains Concept B as a component, element, or part. B is nested within A.

When to Use:

  • B is a named component of framework A
  • B is a step in process A
  • B is a required element of A
  • A explicitly includes B in its definition

Component Types:

  • Prerequisite: B must exist before A
  • Component: B is a part of A
  • Variant: B is a subtype of A

Examples:

  • "Hero's journey" COMPOSED_OF→ "Departure"
  • "Hero's journey" COMPOSED_OF→ "Initiation"
  • "Hero's journey" COMPOSED_OF→ "Return"
  • "Trust" COMPOSED_OF→ "Competence trust" (variant)
  • "Trust" COMPOSED_OF→ "Character trust" (variant)

Not COMPOSED_OF:

  • B merely relates to A (use INFLUENCES or other)
  • B is evidence for A (use SUPPORTS)
  • B causes A (use INFLUENCES)

DERIVES_FROM

Definition: Concept A is logically derived from, emerges from, or is a consequence of Concept B. A exists because of B.

When to Use:

  • A is a logical conclusion from B
  • A emerges from B in a system
  • A is an application of B to a context
  • A is a specialization of B

Derivation Types:

  • Logical conclusion: A follows from B by reasoning
  • Emergent property: A arises from B in complex systems
  • Synthesis: A combines multiple sources including B

Examples:

  • "Community guidelines" DERIVES_FROM→ "Shared values"
  • "Specific onboarding tactics" DERIVES_FROM→ "Engagement theory"
  • "Leadership practices" DERIVES_FROM→ "Trust principles"

Not DERIVES_FROM:

  • B influences A (that's INFLUENCES)
  • A is part of B (that's the reverse - B COMPOSED_OF A)
  • A is evidence for B (that's the reverse - A SUPPORTS B)

Creating Relationships

When to Create a Relationship

Create a relationship when:

  1. The connection is explicitly stated in the source
  2. The connection is strongly implied by context
  3. The connection would be useful for synthesis and querying

Do NOT create relationships:

  • Based on mere co-occurrence in text
  • When the connection is trivial or obvious
  • When evidence for the connection is weak

Relationship Density

Aim for:

  • Every concept linked to at least 1-2 others
  • Key concepts (hubs) linked to many others
  • Both within-book and cross-book relationships

Avoid:

  • Orphan concepts with no relationships
  • Over-connected concepts (>10 relationships may indicate over-extraction)
  • Circular relationships without clear semantics

Relationship Confidence

Track confidence in relationships:

  • High (0.8-1.0): Explicitly stated in source
  • Medium (0.5-0.7): Strongly implied or inferred
  • Low (0.2-0.4): Reasonable inference, less certain

Flag low-confidence relationships for human review.


Cross-Book Relationships

When analyzing multiple books, additional patterns emerge:

Agreement

Both authors say A INFLUENCES B → Stronger confidence

Complementarity

Author 1 says A INFLUENCES B Author 2 says B INFLUENCES C Combined: A → B → C pathway

Contradiction

Author 1 says A SUPPORTS claim X Author 2 says B CONTRADICTS claim X Flag for synthesis discussion

Extension

Author 1 defines framework F Author 2 applies F to new context DERIVES_FROM relationship across books


Relationship Examples by Domain

Community Building

  • "Trust" INFLUENCES→ "Participation" (+0.8)
  • "Shared identity" INFLUENCES→ "Retention" (+0.7)
  • "Explicit guidelines" SUPPORTS→ "Safe participation"
  • "Top-down control" CONTRADICTS↔ "Self-organization"
  • "Engagement ladder" COMPOSED_OF→ "Lurker stage"

Learning

  • "Active recall" INFLUENCES→ "Retention" (+0.9)
  • "Spacing effect" SUPPORTS→ "Spaced repetition works"
  • "Massed practice" CONTRADICTS↔ "Distributed practice"
  • "Learning pyramid" COMPOSED_OF→ "Reading (10%)"
  • "Retrieval practice" DERIVES_FROM→ "Testing effect research"

Leadership

  • "Psychological safety" INFLUENCES→ "Team performance" (+0.8)
  • "Google Project Aristotle" SUPPORTS→ "Safety matters"
  • "Servant leadership" CONTRADICTS↔ "Command-and-control"
  • "Situational leadership" COMPOSED_OF→ "Directing style"
  • "Specific feedback tactics" DERIVES_FROM→ "Growth mindset theory"

Source: SKILL.md on GitHub

2 warnings14d5 checks · Risk SAFE
  • Gen Agent Trust Hub14d

    The skill is safe for parsing and analyzing local ebook collections using Deno-based scripts and LLM prompts. It presents a low-risk surface for indirect prompt injection if malicious content within analyzed ebooks attempts to influence the agent during extraction or classification phases.

  • Socket14d

    No alerts

  • Snyk14d

    Risk: MEDIUM · 1 issue

  • Runlayer7mo

    22/22 files flagged

  • ZeroLeaks5mo

    Score: 93/100 · 2 sections analyzed

Signed by skilld at 7eca2c9. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 2 months ago.

Dormantupdated 8 months ago
Other metadata
metadata
{
  "author": "jwynia",
  "version": "2.1",
  "domain": "research",
  "cluster": "media-meta-analysis",
  "type": "orchestrator",
  "mode": "diagnostic+generative",
  "maturity": "working",
  "maturity_score": 14
}

README badge

README badge for jwynia/agent-skills/ebook-analysis