Skip to content
SkillMixed1,504 tokens in use

review-walkthrough

Turns the current branch into a self-contained HTML walkthrough: the diff split into ordered steps, each with an explanation and review findings anchored to lines.

8,320 stars on the repository CC-BY-SA-4.0trailofbits/skills · plugins/review-walkthrough/skills/review-walkthrough
/plugin marketplace add trailofbits/skills
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing review-walkthrough costs

22tokens idle
1,504tokens in use

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Counts are approximate, made offline with cl100k_base.

Model dependence Mixed

Script-ledMixedModel-led

Scripts do part of the work and the model's judgement does the rest.

This describes the skill's structure, how much of it is scripts. It does not say how well it works. Scripts: scripts/render_walkthrough.py.

Touches

shelldetectednetworkdetectedwrites filesdetectedgitdetected

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Fair
  • Frontmatter8/15
  • Structure6/25
  • Examples6/15
  • Output15/15
  • Size15/15
  • Resources3/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Excellent
  • Runs shell commands−5

    does this on the machine, and the frontmatter does not say so

    SKILL.md, line 105
  • Writes files−8

    does this on the machine, and the frontmatter does not say so

    scripts/render_walkthrough.py, line 243

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Good
  • Development activity100
  • Maintenance52
  • Community64
  • Maturity32
  • Momentum40

review-walkthrough captures the full diff of the current branch against its base and builds a single HTML page for reading it. Files are grouped into steps by concept, with definitions before their consumers and tests beside their subjects, and every step carries an explanation and a list of findings with severity.

It runs only when you call it by name. When the branch belongs to a GitHub pull request, the page lets you pick findings and copy a gh api command for inline review comments. The page never posts anything itself.

What You Get

  • A standalone HTML file with the branch diff split into ordered, explained steps
  • Findings per step with high, medium or low severity, anchored to a file and line on the left or right side of the diff
  • A copyable gh api command for selected findings when the branch has a matching GitHub PR
  • A note on any uncommitted work left out of the walkthrough

Common Use Cases

  • Reading a large pull request in a sensible order before approving it
  • Preparing a branch for reviewers with an explanation of each change
  • Sorting out review comments and exporting the chosen ones to a PR

What Makes It Unique

The reading order follows comprehension, not commit order, while the renderer checks that no diff is rewritten, dropped or repeated. Findings must anchor to real lines in the displayed diff, and the PR export copies a command for you to run yourself, so nothing is posted to GitHub automatically. The skill is read-only: it does not run the project’s tests or install its dependencies.

Under The Hood

The agent runs git commands (rev-parse, merge-base, diff, and a symbolic-ref lookup) to capture the patch, and may query GitHub PR metadata when it is available. It writes a JSON file and a patch file to temporary locations outside the checkout, then runs the bundled Python renderer through uv. The renderer calls git apply to parse paths, validates the data and writes the HTML file. The agent then opens the file in a browser with open or xdg-open when one exists. The generated page itself only builds a gh api command for you to copy; it does not run it. The skill can be invoked only by name.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers