collect-github-stats

name	collect-github-stats
description	Collect GitHub repo health/usage stats into merged JSON.
disable-model-invocation	true
user-invocable	true

collect-github-stats

Runs one script: fetch GitHub stats for a repo, merge into docs/github-stats.json.

Run (MUST)

python3 <skill_dir>/scripts/collect_stats.py [--repo OWNER/NAME] [--out PATH] [--no-enrich]

Defaults: --repo griddynamics/rosetta, --out docs/github-stats.json. MUST run from repo root (or pass absolute --out). --no-enrich skips company/location lookups.

CRITICAL

• NEVER DELETE OR OVERWRITE docs/github-stats.json. Script APPEND-MERGES. Traffic is a 14-DAY ROLLING WINDOW — deleted history is GONE FOREVER, unrecoverable from the API.
• PUSH ACCESS REQUIRED. No push → traffic 403 → script FAILS LOUD with fix steps. MUST NOT accept a partial file.

Rules (MoSCoW)

• MUST have gh authenticated, repo scope, push access to the target repo.
• SHOULD relay the script's post-run summary to the user.
• COULD normalize company variants when REPORTING; MUST NOT mutate stored raw values.
• WON'T fabricate cloner identity (cloners are ANONYMOUS — counts only), store PII/secrets.

Lessons learned (keep updating, first line is template, follow )

• <concise: what happened, why, root cause, reasoning>.