Fiction Linter

Fiction Linter is a VS Code extension designed to act as a Semantic Physics Engine (SPE) for your fiction writing. It helps you avoid "First Probability" prose—generic, clichéd, or AI-sounding text—by rigorously checking your manuscript against a set of constraints.

Features

• Pattern Linter: Detects and flags:
  • Cliches: Overused phrases (e.g., "shiver down spine", "released a breath").
  • AI Tells: Structural patterns common in AI writing (e.g., "It is worth noting", "couldn't help but").
  • Weak Phrasing: Vague descriptors or lack of sensory precision.
• Name & Place Validator:
  • Checks character names against a blocklist of overused AI names.
  • Flags generic or clichéd place names (e.g., "Willow Creek", "Ironpeak").

Visual Guide

The extension uses standard VS Code diagnostic colors to indicate the severity of the issue:

• Red Squiggles (Error): Strict Bans & AI Patterns.

  • Examples: "shiver down his spine", "It is worth noting that", "walls I've built".
  • Action: These must be removed or rewritten.

• Yellow Squiggles (Warning): Cliches, Tells, & Names.

  • Examples: "released a breath", "felt sad", "Kael", "Willow Creek".
  • Action: Avoid these unless strictly necessary or contextually justified.

• Blue Squiggles (Info): Weak Phrasing & Style.

  • Examples: "very", "just", "practiced ease".
  • Action: Consider strengthening the prose with more specific descriptors.

Configuration

Zero Config: By default, the extension comes with a bundled set of standard Semantic Physics Engine rules (Cliches, Names, etc.). You don't need to do anything to start linting!

Advanced (Custom Rules): If you have your own customized SPE definitions (YAML files), you can point the extension to them:

1. Open VS Code Settings (Cmd+,).
2. Search for Fiction Linter.
3. Set the Spe Path to the absolute path of your directory containing the .yaml files.
   • Example: /Users/yourname/projects/my-spe-config

The extension looks for:

• cliche_collider.yaml
• name_collider.yaml
• place_collider.yaml
• line_editing_protocol.yaml

AI Features

This extension includes AI-powered tools to help you fix prose issues:

• Fix with AI: Click "Quick Fix" on any squiggly line to have AI rewrite the sentence.
• Deep Scan (AI Scan):
  • Button: Click the $(search) AI Scan button in the status bar (bottom left).
  • Scope:
    • Chapter Mode: If your cursor is inside a chapter (marked by ## Chapter X), it scans only that chapter.
    • Full Mode: If no characters are detected, it scans the entire document.
  • Progressive: Scans in background chunks (default 5 paragraphs) so you can keep working.
  • Cancel: Click the button again ($(sync~spin) Stop Scan) to cancel immediately.

Configuration: The extension attempts to use available VS Code Copilot models first. If none are found, you can configure a direct connection to OpenAI:

1. fiction-linter.openAiKey: Your OpenAI API Key.
2. fiction-linter.modelName: (Optional) Model to use (default: gpt-4).
3. fiction-linter.apiBaseUrl: (Optional) Custom endpoint (default: OpenAI).
4. fiction-linter.autoAiScanChunkSize: Number of paragraphs to scan at once (default: 5). Increase for faster scans if your API limits allow.

Requirements

• VS Code 1.100.0 or higher.
• A local copy of the Semantic Physics Engine data files.

Release Notes

0.1.5

• AI Scan Button: Added status bar button to trigger progressive AI scanning.
• Chapter Detection: Scans are now scoped to the current chapter.
• Bug Fixes: Improved Markdown activation and stability.

0.0.1

• Initial release.
• Support for Cliche Collider, Name Collider, and Place Collider.

Enjoy writing high-entropy fiction!