Does this skill work with all programming languages?

Yes, it supports a wide range of languages by automatically detecting the correct comment character (such as # or //) and utilizes YAML frontmatter specifically for Markdown files.

Can Claude mark a file as human-reviewed using this skill?

No, the skill explicitly prevents AI agents from setting 'human-reviewed' to true. Only human developers can manually verify and update this status to ensure code quality.

How do I see which files in my project haven't been reviewed yet?

You can use the 'scan' command which generates a comprehensive report of your repository, grouping files by directory and showing counts for stamped, reviewed, and unstamped files.

What happens when an AI modifies a file that was already reviewed?

The skill automatically resets the 'human-reviewed' flag to false whenever the AI agent edits a file, signaling that the new changes require a fresh human audit.

Can I prevent the AI from seeing or editing specific files?

Yes, by setting the 'access' field to 'read' (read-only) or 'hidden' (no access), you can strictly control the agent's permissions on a per-file basis.

AI Frontmatter & Provenance

Name: AI Frontmatter & Provenance
Author: a3lem

bya3lem

0•

开发者工具

Manages AI provenance metadata blocks to track human review status, rule references, and file-level access permissions.

The Frontmatter skill enables developers to maintain a clear audit trail of AI-generated content by embedding metadata directly into source files. It facilitates a safer 'human-in-the-loop' workflow by tracking which files have been human-reviewed, enforcing file-level access controls to prevent unauthorized AI modifications, and mapping specific development rules to individual code blocks. Whether you are managing a large-scale migration or daily AI-assisted coding, this skill ensures transparency and control over how AI agents interact with your repository, automatically invalidating review statuses whenever the AI modifies a file.

主要功能

01Recursive directory scanning to report metadata coverage and review status

02Enforces file-level access permissions including read, write, and hidden modes

03Multi-language support with automatic comment syntax detection for various extensions

04Automatic reset of review status upon any AI-driven file modifications

05Automated '/// ai' metadata stamping for provenance tracking

060 GitHub stars

使用场景

01Protecting sensitive or legacy files from AI modifications via read-only tags

02Auditing repositories to distinguish between human-verified code and raw AI output

03Context-aware development by linking specific style guides to individual source files

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add a3lem/my-claude-plugins frontmatter

For use in Claude.ai and ChatGPT

主要功能

01Recursive directory scanning to report metadata coverage and review status

02Enforces file-level access permissions including read, write, and hidden modes

03Multi-language support with automatic comment syntax detection for various extensions

04Automatic reset of review status upon any AI-driven file modifications

05Automated '/// ai' metadata stamping for provenance tracking

060 GitHub stars

使用场景

01Protecting sensitive or legacy files from AI modifications via read-only tags

02Auditing repositories to distinguish between human-verified code and raw AI output

03Context-aware development by linking specific style guides to individual source files