doc-review_skill

This skill helps you review documents thoroughly by generating structured feedback and improvement suggestions in a consistent markdown format.
  • Shell

0

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill shotaiuchi/dotclaude --skill doc-review

  • SKILL.md7.5 KB

Overview

This skill creates structured review reports for document files and writes them to docs/reviews/<path>.<filename>.md. It supports single-file and parallel multi-file processing via a doc-reviewer sub-agent, with automatic glob expansion and batched concurrency for large sets. Reviews are written in Japanese, include prioritized improvement suggestions, and preserve a clear output summary.

How this skill works

The skill parses the input file path(s) or glob pattern, validates file existence, and decides processing mode based on the number of files (single, async parallel, or batched parallel). For each file it invokes the doc-reviewer sub-agent with standardized placeholders (filename, date, file_path) and instructions to load a review template, evaluate purpose/clarity/completeness/accuracy, and produce a prioritized review written to docs/reviews/<path>.<filename>.md. When multiple files are processed, all Task calls are launched in one assistant message with run_in_background: true to ensure true parallel execution; results are later collected and summarized.

When to use it

  • You need a concise, consistent review for one or many documentation files.
  • Batch-reviewing a docs directory before a release or documentation cleanup.
  • Generating reviewer output to guide edits: issues, locations, and concrete suggestions.
  • When you want parallel processing to speed reviews across many files.
  • Retrying failed reviews selectively without reprocessing successful files.

Best practices

  • Provide explicit file paths or glob patterns (e.g., docs/*.md) to target sets of files.
  • For multiple files, ensure all Task invocations are sent in a single assistant message with run_in_background: true.
  • Keep batches to at most 5 concurrent tasks; the skill enforces MAX_PARALLEL = 5.
  • Review generated docs/reviews/*.md promptly and apply high-priority suggestions first.
  • Retry only the failed files using the same command to avoid re-running completed reviews.

Example use cases

  • Review a single README: /doc-review docs/README.md
  • Review a set of guides before release: /doc-review docs/*.md
  • Audit technical docs for clarity and correctness across a subfolder: /doc-review dotclaude/commands/wf*.md
  • Run large-scale reviews in batches: /doc-review docs/**/*.md (skill will batch by 5)

FAQ

All review files are written in Japanese regardless of the document language.

How does parallel processing work for many files?

Files are launched as background Task calls in groups of up to 5; the skill collects TaskOutput results and summarizes succeeded and failed files.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational