Skip to content

Stats

The stats command generates a self-contained HTML report describing pull request activity for one or more repositories during a period.

Usage

ghx stats [REPO...] [flags]

Repositories are positional [HOST/]OWNER/REPO arguments. When none are given, the repository is resolved from --repo or the current directory's git remote.

# Report for the current repository over May 2026
ghx stats --from 2026-05-01 --to 2026-05-31

# Report for two repositories, filtered to specific people
ghx stats acme/widget acme/gadget --from 2026-05-01 --to 2026-05-31 \
    --author alice,carol --reviewer alice,bob --output may-report.html

Run ghx cache first so the report covers full history; stats serves data from the local cache and only falls back to the GitHub API (capped at 1000 PRs) when nothing is cached.

Flags

Flag Description
--author, -A Only count PRs authored by these users (comma-separated)
--reviewer Only count comments by these reviewers (comma-separated)
--from Period start (YYYY-MM-DD or RFC3339)
--to Period end, inclusive (YYYY-MM-DD or RFC3339)
--state, -s Filter by PR state: open, closed, merged, all (default all)
--output, -o Output HTML file path (default stats-report.html)
--include-bots Include bot accounts (*[[bot]]) as reviewers
--top Number of pull requests per notable-PRs list (default 5)
--list-prs Include the table of individual pull requests (can be large)

Report contents

  • Summary cards: total PRs, merged/open/closed counts, author and reviewer counts, median/average time to first comment, median time to merge, and the total number of submitted reviews.
  • Monthly trend chart: PRs opened and merged per month plus the median time from opening to merge (Chart.js, loaded from the jsDelivr CDN).
  • Activity by author: one bar chart per PR author showing how many PRs they opened each month (by creation date) alongside how many they had merged or closed that month (Chart.js, loaded from the jsDelivr CDN).
  • Merge speed (merged PRs only): how many merged PRs landed within 1 hour, 1 day, and 1 week of being opened, each with its share of merged PRs, plus the p50/p90/p99 percentiles of the time from opening to merge. The same statistics are also broken down per author in a "Merge speed by author" table.
  • Merge speed over time: line charts of the within-1h/1d/1w shares and the p50/p90/p99 of time to merge, grouped by merge date, with daily, weekly, and monthly granularity buttons, a Cumulative toggle that aggregates the PRs merged up to each point instead of only that bucket's PRs, a Window field bounding how far back each cumulative point looks (default 28d, units d/w/m, empty for all past), and From/To date fields limiting the displayed period (Chart.js, loaded from the jsDelivr CDN).
  • Author × reviewer matrix: authors as rows, reviewers as columns. Each cell shows how many PRs the combination appears in, the share of that reviewer's total reviewed PRs (column headers show the reviewer total in parentheses), plus the average and median time from PR creation to that reviewer's first comment. Cells are shaded by count, and a (no comments) column tracks PRs that nobody reviewed. Two dropdown filters above the table (Tom Select, loaded from the jsDelivr CDN) narrow the matrix to a subset of authors (rows) and reviewers (columns) with type-ahead autocompletion and multi-select; the (no comments) and Total columns always stay visible, and a hint reports how many authors and reviewers are shown.
  • Per-author summary: PRs authored, merged count, and time to first comment received.
  • Lead time (merged PRs only): per author, the median and average time from opening to merge, plus median time in draft, median time to first review, median time to approve, and median time from approval to merge (first approving review to the merge, counting only PRs approved before they were merged).
  • Contribution: per author, opened/merged/closed counts, merge rate, PRs without any review activity, comments received, total additions/deletions, and the distribution of PR sizes (xs/s/m/l/xl by changed lines: xs <100, s <500, m <1000, l <2000, xl >=2000).
  • Per-reviewer summary: PRs reviewed and time to first comment given.
  • Reviewer engagement: per reviewer, comment and review counts, approvals, changes requested, comment-only reviews, median time from PR opening to the reviewer's first review, and median time from a review request (and re-request) to that reviewer's review.
  • PR size vs merge time: per size bucket, how many PRs were opened and merged and the median time to merge, plus a bar chart of the size distribution.
  • Peak activity: a bar chart of when PRs were opened by hour of day.
  • Notable pull requests: the longest PRs awaiting a first review, the longest PRs to merge, and the most discussed PRs (--top entries each).
  • Pull requests in period (only with --list-prs): the full list of matching PRs with links. Off by default since the table can be very large for active repositories.
  • Section sidebar: a fixed, full-height table of contents on the left, with the section currently in view highlighted while scrolling. On narrow screens it is replaced by a drawer opened by swiping right from the left edge or tapping the fixed ☰ button (dismissed by the ✕ button, swiping left, tapping outside, or pressing Escape).

Reviewers are derived from PR comments and reviews written by someone other than the PR author; bot accounts are excluded unless --include-bots is set.

Cache data requirements

The lead time, contribution, engagement, and size sections rely on review, timeline, and change-size data that older cache entries do not contain. When the report detects PRs cached before this data was collected it shows a notice; run ghx cache --force for the affected repositories to collect the full data set. Sections still render with a - for PRs where the underlying events were never recorded (for example PRs that never received a formal review).

Themes

The report ships with light and dark themes. It defaults to the browser's prefers-color-scheme, and the theme toggle button in the top-right corner switches between them. The choice is remembered per browser via localStorage, so re-opening the report keeps the selected theme.