Contributor stats using gh CLI tool

[wjt]

I wanted to get some rough statistics on the number of people contributing to Threadbare in various ways. The easiest way I found to get the numbers I wanted was to use the gh command-line tool to query issues; tell it to present the result in JSON format; and tell it to run that JSON through a jq program.

I want to exclude people who work for Endless Access from these numbers. Unfortunately the query language that GitHub provides for issues and pull requests doesn't support saying "exclude issues assigned to members of this team". So I resorted to a cut-off date (before which more people from Endless Access and our partners were committing actively) plus a list of "except this person" rules that I wrote by hand.

Contributors assigned to resolved issues

gh issue list --limit 1000 -s all \
  --search '-assignee:cassidyjames -assignee:PlayMatters -assignee:wjt -assignee:manuq closed:>2025-06-01 reason:completed is:issue has:assignee' \
  --json assignees \
  --jq 'map(.assignees[].login) | unique | sort | length'
# ⇒ 25

Complications here:

• GitHub Issues can be closed with two reasons: "completed" and "not planned". We only want to count completed issues. Unfortunately when you convert an issue into a Discussion, it is also apparently considered completed. To exclude these I used has:assignee to only select issues that were assigned to someone.
• I can't remember why I had to say -s all rather than -s closed here.

Authors of accepted pull requests

$ gh pr  list --limit 1000 -s merged \
    --search 'merged:>=2025-06-01 -author:cassidyjames -author:PlayMatters -author:hydrolet -author:wjt -author:manuq -author:app/dependabot -author:jgbourque' \
    --json author \
    --jq 'map(.author.login) | unique | length'
# ⇒ 16

This number is not the same as the number of contributors who resolved issues, because:

1. Not all issues need to result in a pull request to be resolved. For example:
   • Playtest all quests & file aspect ratio bugs #1083 would be resolved by reporting on the playtesting & filing 1+ more issues
   • Some issues are resolved by providing an asset, sketched level design, concept art, etc.: integrating these is tracked separately.
2. A pull request can have contributions from more than one person, but it is not easy to access this data:
   • PRs can only have one creator
   • PRs do have an Assignees field (like Issues) but only maintainers can assign people to a PR; and even then only people who have commented on the PR; and this is manual
   • You can dig down into the commits to find their (potentially multiple) authors but we can't quickly do this with gh

Contributors assigned to open issues

You'd think this would be easy, but it's not: has:assignee does not seem to work in gh's --search argument. However you can just look at this on the web:

https://github.com/endlessm/threadbare/issues?q=-assignee%3Acassidyjames%20-assignee%3Awjt%20-assignee%3Amanuq%20is%3Aopen%20is%3Aissue%20has%3Aassignee

Average XP earned

Many of our issues are tagged with an amount of XP earned by completing them. This is input into another system that we haven't built yet, but we want to take a look at the average so far. So with this jq program saved as xp.jq:

# Parse XP out of comment bodies
map({assignees: .assignees|map(.login), "xp": .body | capture("XP Summary \\(total (?<xp>\\d+)\\)") | .xp}) |

# Explode multiple assignees to one row per assignee
map(
  .xp as $xp |
  .assignees[] |
  {assignee: ., xp: ($xp | tonumber)}
) |

# Group by assignee and get the total per assignee
group_by(.assignee) |
map({assignee: .[0].assignee, total_xp: map(.xp) | add}) |

# Now average that
map(.total_xp) | add / length

We can then filter the output of a query through it:

gh issue list --limit 1000 -s all \
    --search '-assignee:cassidyjames -assignee:PlayMatters -assignee:wjt -assignee:manuq closed:>2025-06-01 reason:completed is:issue has:assignee'
    --json assignees,body |
    jq -f xp.jq
# ⇒ 318