Agent Skill · Apollo Config
apollo-contributor-promotion-review
Review Apollo community contributors for quarterly promotion readiness across the apolloconfig organization. Use when Codex needs to scan the last 100 days of contributor activity, rank top contributors, inspect their historical PR/review/community evidence, and produce an internal report that recommends member invitation, committer discussion, PMC discussion, or continued observation.
Skill body
Apollo Contributor Promotion Review
Use this skill to run a quarterly Apollo contributor promotion review. It scans recent activity across apolloconfig, ranks the top contributors, backfills historical evidence, and produces an internal report for governance decisions.
Path notes:
- Assume
SKILL_ROOTpoints to this skill directory. - Example:
export SKILL_ROOT=/path/to/apollo-contributor-promotion-review
Defaults
- organization:
apolloconfig - lookback window:
100days - ranking size: top
10 - primary data sources:
- GitHub GraphQL search for recent PRs and issues
- GitHub REST search for top-candidate historical evidence
apolloconfig/apollodocs/en/community/team.mdfor PMC / committer detectionapolloconfig/apollo-communitymembership request issues for member detection
- default policy file:
$SKILL_ROOT/references/review-policy.example.json - optional role override file:
--role-overrides /path/to/role-overrides.json
Workflow
- Review the policy inputs.
- Use the default policy unless the user explicitly wants a different repo scope, lookback window, or score weights.
- Use a role override file when the public GitHub data is known to be incomplete for members, committers, PMC, aliases, or ignore rules.
- Run the scan.
python3 "$SKILL_ROOT/scripts/contributor_promotion_review.py" \ scan \ --policy-file "$SKILL_ROOT/references/review-policy.example.json" \ --pretty > /path/to/scan.json - Render the internal report.
python3 "$SKILL_ROOT/scripts/contributor_promotion_review.py" \ summarize \ --scan-file /path/to/scan.json - Review the recommendations manually before taking action.
Membermeans the evidence is strong enough to consider sending a membership invitation flow, but 2FA and sponsorship remain manual checks.Committer discussionmeans the candidate looks ready for a PMC nomination discussion, not that the promotion should happen automatically.Candidate for PMC discussionmeans the candidate looks strong enough for internal PMC consideration, not that the vote has happened.Continue observingmeans the skill found useful evidence but not enough for a promotion recommendation.
Output Rules
- Prefer the markdown report for user-facing output.
- Keep absolute dates in the report so the review window is explicit.
- Show 3 to 5 evidence links per ranked contributor.
- Preserve the conservative recommendation style:
- only recommend
Memberwhen the evidence is clearly above threshold - only recommend
Committer discussionfor recognized members - only recommend
Candidate for PMC discussionfor recognized committers - route everyone else to
Continue observing
- only recommend
Safety Rules
- Never auto-post issue comments, PR comments, or discussions from this skill.
- Never create membership requests, committer nominations, PMC nominations, or team-list PRs unless the user asks for a separate follow-up task.
- Treat 2FA, internal voting, contribution difficulty, and private governance context as manual checks.
- If GitHub data is incomplete or ambiguous, keep the recommendation conservative and list the missing evidence in
blockers.
Resources
scripts/
contributor_promotion_review.py: fetches GitHub data, aggregates recent contributions, enriches top candidates with historical evidence, and renders a markdown report.
references/
review-policy.example.json: default scoring, scope, and threshold policy for quarterly reviews.role-overrides.example.json: optional manual overrides for known members, committers, PMC, aliases, or ignored logins.