Agent-Skills.md

Agent Skills: deviation-impact-analysis

Turn "AI draft -> author edits" diffs into structured deviation items, an author-intent prompt, and a concrete propagation plan.

UncategorizedID: aevatarAI/aevatar-agent-framework/deviation-impact-analysis
117

Install this agent skill to your local

pnpm dlx add-skill https://github.com/aevatarAI/aevatar-agent-framework/tree/HEAD/experimental/novel/agent_skills/deviation-impact-analysis

Skill Files

Browse the full folder contents for deviation-impact-analysis.

Download Skill

Loading file tree…

experimental/novel/agent_skills/deviation-impact-analysis/SKILL.md

Skill Metadata

Name
deviation-impact-analysis
Description
Turn "AI draft -> author edits" diffs into structured deviation items, an author-intent prompt, and a concrete propagation plan.

When to use

  • After author edits a chapter draft (especially via external editor like Cursor)
  • Before auto-continue to future chapters

Outputs

  • deviation_report.md (human readable summary)
  • deviation_prompt.md (one-click prompt/instructions)
  • change_impact_report.md (what future outline/chapters/canon/ledger will be affected)
  • backup_options_report.md (conservative vs aggressive vs branch)

Structured deviation categories (must classify)

  • plot / character / tone / info density / canon setting / timeline / style / pacing

Procedure

  1. Extract diff signals and cluster them into 5-20 deviation items (avoid noise).
  2. Summarize author intent in a short instruction that can drive future rewriting.
  3. Scan future outline/chapters/canon/ledger:
    • list concrete touch points (file/section) that need change
  4. Produce 2-3 backup routes and recommend one.