Claude Plugin Manifest for Humanizer: Complete Configuration Guide

The Claude plugin manifest for Humanizer is located at .claude-plugin/plugin.json and defines the plugin metadata including name, version, description, and a pointer to the skill file that contains the rewrite logic.

The blader/humanizer repository includes a Claude-compatible plugin manifest that enables seamless integration with Anthropic's Claude AI assistant. This JSON configuration file tells Claude how to load and execute the Humanizer skill for rewriting AI-generated text into natural language. Understanding this manifest structure is essential for developers looking to deploy or customize the Humanizer plugin.

Manifest Location and Structure

The Claude plugin manifest resides at .claude-plugin/plugin.json in the repository root. This file follows Claude's plugin specification format, providing the metadata required for the Claude client to identify, validate, and initialize the plugin.

The manifest uses standard JSON structure without YAML frontmatter. All paths referenced within the manifest are relative to the repository root, ensuring portability across different installation environments.

Core Configuration Fields

The plugin.json file contains several required fields that define the plugin's identity and capabilities:

• name: "humanizer" — The unique identifier used by Claude to reference the plugin internally.
• description: "Rewrite AI-sounding text so it reads naturally without changing what it says." — A user-facing explanation of the plugin's functionality.
• version: "3.0.0" — Semantic version tracking for the plugin release.
• author: An object containing the plugin author's details, including name, email, and url fields.
• skill: "SKILL.md" — The relative path to the skill file containing the actual prompt logic and text transformation patterns.

Claude's plugin specification requires at minimum the name, description, version, and author fields, plus a valid skill pointer that directs the loader to the implementation file.

Skill File Integration

The skill field in the manifest points to SKILL.md at the repository root. This separation of concerns allows the manifest to handle metadata while the skill file contains the implementation details, prompt templates, and rewrite patterns.

When Claude loads the Humanizer plugin, it reads SKILL.md as instructed by the manifest's skill entry. This file contains the core logic that transforms AI-sounding text into natural prose without altering the underlying meaning.

Supporting Files and Validation

The repository includes additional configuration files that support the Claude plugin ecosystem:

• .claude-plugin/marketplace.json: Optional metadata for publishing the plugin on Claude's marketplace.
• agents/openai.yaml: Example configuration demonstrating how to use Humanizer with OpenAI-compatible agents.
• scripts/validate-package.py: Python validation script that checks the integrity and format compliance of both the manifest and skill files.

These files work together to ensure the plugin meets Claude's technical requirements before deployment.

Example Manifest Configuration

Here is the complete JSON structure found in the Humanizer repository:

{
  "name": "humanizer",
  "description": "Rewrite AI-sounding text so it reads naturally without changing what it says.",
  "version": "3.0.0",
  "author": {
    "name": "blader",
    "email": "you@example.com",
    "url": "https://github.com/blader"
  },
  "skill": "SKILL.md"
}

When integrated, Claude parses this manifest to register the plugin, then loads the referenced skill file to make Humanizer's text transformation capabilities available to the model.

Summary

• The Claude plugin manifest for Humanizer is located at .claude-plugin/plugin.json in the blader/humanizer repository.
• Required fields include name ("humanizer"), description, version ("3.0.0"), author object, and skill pointer.
• The skill field references SKILL.md, which contains the actual prompt logic and rewrite patterns.
• Supporting files include marketplace.json for publishing metadata and validate-package.py for integrity checking.
• The manifest follows Claude's plugin specification format for seamless integration with the Claude AI assistant.

Frequently Asked Questions

Where is the Claude plugin manifest located in the Humanizer repository?

The manifest is located at .claude-plugin/plugin.json relative to the repository root. This file contains the JSON metadata that Claude uses to identify and load the Humanizer skill.

What fields are required in the Humanizer plugin manifest?

According to the source code, the manifest must include name (set to "humanizer"), description (explaining the rewrite functionality), version (currently "3.0.0"), an author object with contact details, and a skill field pointing to the implementation file.

What is the purpose of the SKILL.md file referenced in the manifest?

SKILL.md contains the actual Humanizer prompt logic and text transformation patterns. The manifest's skill field tells Claude where to find these instructions, separating metadata from implementation.

How can I validate the Humanizer plugin manifest before deployment?

Use the scripts/validate-package.py script included in the repository. This Python utility checks the manifest and skill files for format compliance and structural integrity according to Claude's plugin specification.

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