accelint-readme-writer_skill

Overview

This skill analyzes a project's files from a chosen directory and generates or updates a README.md that matches the codebase. It finds public APIs, examples, scripts, and dependencies, then produces clear, copy-pasteable documentation and practical examples. It also flags missing or stale sections and suggests concrete edits.

How this skill works

The skill recursively parses the code starting at the README location to identify entry points, exported symbols, example files, and package metadata. It compares discovered functionality against the existing documentation to find gaps, outdated examples, and wrong commands. Finally it generates a structured Markdown document or a patch with specific changes, including installation commands detected from lockfiles and runnable code blocks.

When to use it

• Creating a README for a new package or subpackage
• Updating docs after refactors, exports changed, or API renamed
• Auditing docs for completeness before release
• Converting sparse or legacy docs into modern, example-driven guides
• When asked to "document this package" or to write docs for a monorepo subdirectory

Best practices

• Run the skill from the directory where the README should live so scope is correct
• Confirm package manager based on lockfiles before applying install commands
• Review suggested examples and run them locally before committing
• Use the humanizer step to remove AI writing artifacts and ensure natural tone
• Accept changes incrementally: review suggested diffs rather than auto-applying everything

Example use cases

• Generate a new README for a freshly-bootstrapped library with index exports and tests
• Update a package README after moving exports between files or renaming functions
• Add a quickstart and copy-paste examples for a CLI tool, with correct install command detected from lockfile
• Audit a monorepo package folder and produce a focused README describing its public API and scripts
• Convert inline examples from tests into runnable snippets in the documentation