🟡 The star drift detector

You get: proof that your published numbers are stale. Pairs with: humility.

for repo in anthropics/claude-code anomalyco/opencode openai/codex; do
  echo "$repo: $(tabstack extract json "https://github.com/$repo" 
    --schema '{"type":"object","properties":{"stars":{"type":"number"}}}' 
    | jq .stars)"
done

What it does

Extracts the current star count from each GitHub repo as a number. Run it against repos you’ve mentioned in a blog post, README, or talk — the output tells you how stale your numbers are.

What it caught in practice

This recipe was used to fact-check a post on agentic harnesses. After 11 weeks of drift:

Two of the ten tools in that post had also been renamed. The follow-up post was written from this diff.

Structured comparison

# Build a before/after table
repos=("anthropics/claude-code" "openai/codex")
claimed=(83000 67700)

for i in "${!repos[@]}"; do
  repo="${repos[$i]}"
  live=$(tabstack extract json "https://github.com/$repo" 
    --schema '{"type":"object","properties":{"stars":{"type":"number"}}}' 
    | jq .stars)
  drift=$(( live - claimed[$i] ))
  echo "$repo: claimed ${claimed[$i]}, live $live, drift +$drift"
done

Beyond GitHub stars

The same pattern works for any numeric stat on any page:

# npm download counts
tabstack extract json "https://www.npmjs.com/package/zod" 
  --schema '{"type":"object","properties":{"weeklyDownloads":{"type":"number"}}}'

# Product Hunt upvotes
tabstack extract json "https://www.producthunt.com/products/tabstack/launches" 
  --schema '{"type":"object","properties":{"upvotes":{"type":"number"}}}'

Add it to your writing workflow

Before publishing, run this against every statistic you cite. Exit non-zero if any number is more than 20% off from what you wrote:

tabstack extract json "https://github.com/some/repo" 
  --schema '{"type":"object","properties":{"stars":{"type":"number"}}}' 
  | jq -e '.stars > 50000' || echo "WARNING: star count may be outdated"