Index of every page on this site: llms.txt
All skills
Run it yourself

Use when a change affects tracked screenshots, captured views, or other user-facing visuals, before committing regenerated screenshot assets, or when generating, reviewing, or promoting images from a repository's deterministic capture lane. Covers routing to the repository's own screenshot skill, keeping screenshot generation and visual approval as separate gates, privacy of captured data, image classes and review strength, the complete-set review process, acceptance criteria, promotion safety, and verifying the rendered product.

  • 1 file
  • 4.1 KB
  • Updated 3 weeks ago
  • GitHub
Use this Skill: https://skilld.dev/gh/bydradev/gh-repo-bootstrapper/screenshot-review

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

SKILL.md

≈139 tokens for metadata: the name and description. ≈907 when used: this file.

Description uses 6.9% of example budget

Before choosing a Skill, your Agent reads its name and description. All available Skills share that space.

  • A shorter description leaves more room for other Skills. This entry exceeds our 1% size suggestion.

In our Claude Code example, all Skill names and descriptions share 8,000 characters. This Skill uses ≈555 characters, or 6.9%.

The 1% threshold is a size suggestion. Longer descriptions can still fit.

Your model, settings, and other Skills decide how much text your Agent can read.

Example settings and source

The example uses a 200k-token context and default Claude Code settings. The count includes the name, description, separators, and when_to_use when present. Codex also counts local file paths.

Skit's source and limits: Codex 0.160.1, Claude Code 2.1.292.

Screenshot review

Screenshot guidance is capability-conditional: do not assume browser or native capture tooling exists in this repository. A successful screenshot-generation workflow proves capture only; it does not approve visual fidelity or privacy.

If ## Project specifics in AGENTS.md names a repository-owned skill, or a section of one, for screenshot or visual-review work, load it first. It holds this repository's capability boundaries and procedures, and baseline pointers where present. Where it is narrower than this skill, it governs.

The hub's screenshot gates in Definition of done apply to every change below; this skill adds the review process.

When the repository has a deterministic capture lane

A capture lane is a repository-provided command that renders fixed, fictional data into a known set of images. These rules apply only when the repository has one; its own screenshot skill names the command, configuration, image set, and current baseline.

Image classes and review strength

Image class Review boundary
Temporary diagnostic artifact, such as a test-failure capture Normally untracked; generation alone is not acceptance
Visual-regression baseline Review the image diff and the behavior behind it in the pull request
Documentation image Review visual quality, truthfulness, accessibility, and privacy
Tracked asset set Review every image as below; the repository skill states whether one pass or two fresh-context passes are required

Review process

  1. Generate a fresh artifact from the intended source commit. For final acceptance, prefer the repository's hosted or CI run over a local one.
  2. Confirm every configured image exists and that no unexpected file was produced.
  3. Inspect every image at full resolution for layout, content, accessibility, and privacy.
  4. Fix product or capture defects in source; never retouch an image to hide them.
  5. Regenerate the complete set and repeat the review after any fix.
  6. For launch-facing images, do a second fresh-context pass against newly generated evidence. A second reviewer is preferred; otherwise separate the passes with unrelated work or a full validation cycle.
  7. Commit accepted images only with the source change that adopts them.

Record the decision, the artifact or run reference, the source commit, the reviewer or review context, the files reviewed, and any unresolved findings.

Acceptance criteria

For every image, confirm:

  • the configured view, viewport, theme, and full-page behavior are as expected;
  • there is no clipping, overflow, overlap, broken truncation, sliced navigation, unreadable wrapping, loading artifact, or development chrome;
  • values are internally consistent, with no undefined, NaN, placeholder copy, contradictory empty state, or claim that fixture data is live;
  • only fictional organizations, people, identifiers, and URLs appear, and no real private name, email address, token-shaped value, internal URL, or unrelated desktop content is visible;
  • text is legible, contrast is credible, focus and hover artifacts are absent, and meaning does not rely on color alone; and
  • a launch-facing image shows the product with useful composition and no error state or development-only presentation.

Promotion safety

Capture into a temporary location and promote only after every configured image succeeds. If promotion replaces files one at a time and fails partway, inspect the diff and the destination set, then rerun the complete capture. Never treat a partial replacement as an accepted set.

Source: SKILL.md on GitHub

No rule matched.

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No third-party reports yet.

Signed by skilld at 20b39bd. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub yesterday.

Activeupdated 3 weeks ago

README badge

README badge for bydradev/gh-repo-bootstrapper/screenshot-review