star-history

name	star-history
description	Show GitHub star history charts. Triggers on star history, star chart, repo popularity, comparing repos by stars.

Star History Skill

Show GitHub repository star history as an inline chart with a link to star-history.com.

Input Parsing

Accept repos in any of these formats:

• owner/repo → use directly
• https://github.com/owner/repo → extract owner/repo
• https://star-history.com/#owner/repo&Date → extract owner/repo
• https://star-history.com/#owner/repo1&owner/repo2&Date → extract multiple repos
• Just name when owner equals repo name (e.g., star-history → star-history/star-history)

Workflow

Step 1: Fetch star data

python3 SKILL_DIR/scripts/fetch_stars.py owner/repo [owner/repo2 ...] 2>/dev/null > /tmp/star_data.json

Output: JSON array of {repo, total_stars, stars: [{date, cumulative}]}. The script samples ~20 evenly-spaced pages from the stargazers API (~40 data points per repo), staying within GitHub's 60 req/hr unauthenticated limit.

Step 2: Render chart (environment-dependent)

Path A — Claude.ai (Visualizer available)

Use visualize:show_widget to render a pure SVG chart inline.

IMPORTANT: Chart.js does NOT work in the Visualizer sandbox. Use hand-drawn SVG only.

You can use render_chart.py --svg-only to generate the SVG, or construct it directly. Either way, pass the raw SVG string as widget_code.

python3 SKILL_DIR/scripts/render_chart.py -i /tmp/star_data.json --svg-only

Path B — Claude Code / Cowork / CLI (no Visualizer)

Generate a standalone HTML file and open it in the browser:

python3 SKILL_DIR/scripts/render_chart.py -i /tmp/star_data.json -o star-history.html --open

The --open flag calls webbrowser.open() to launch it. On headless environments, skip --open and tell the user the file path.

Step 3: Always provide the star-history.com link

Star-history.com URL format:

• Single: https://star-history.com/#owner/repo&Date
• Multiple: https://star-history.com/#owner/repo1&owner/repo2&Date

Include the link:

1. Inside the SVG (the render script adds it automatically as a clickable <a>)
2. In the prose response text as a fallback

Step 4: Handle errors gracefully

• Rate limited (HTTP 403): GitHub's unauthenticated limit is 60 req/hr. Suggest trying later or use the star-history.com link directly.
• Repo not found (HTTP 404): Report the repo doesn't exist or is private.
• Very small repos (<5 stars): Skip the chart, just report the count and provide the star-history.com link.

SVG Chart Design (for Path A manual construction)

If constructing SVG directly instead of using render_chart.py:

• Coordinate system: viewBox="0 0 700 400", chart area (60,30) to (680,330)
• Map dates to X (60–680), cumulative stars to Y (330→30, inverted)
• Horizontal dashed grid lines at Y-axis tick positions
• Y-axis labels with K-suffix (e.g., "10K"). X-axis with month labels (e.g., "Mar 25")
• <polyline> per repo, <polygon> with gradient fill underneath
• End-point dots + total star count annotation
• Legend: colored squares + repo names
• Clickable <a> to star-history.com at bottom-right

Color palette (in order): #534AB7 (purple), #1D9E75 (teal), #D85A30 (coral), #378ADD (blue).

Notes

• For repos with many stars (>10K), total_stars is approximate (pages × 100).
• If GITHUB_TOKEN env var is set, authenticated requests get 5000 req/hr.
• Up to ~4 repos can be compared before unauthenticated rate limits become an issue.