Skip to content

ruvnet.

Python
4.0 KB
Raw Blame

Daily ecosystem metrics

Requested: 2026-10-09. Cadence: daily morning, America/Toronto, managed by ChatGPT automation.

Refresh the ruvnet ecosystem statistics in the public GitHub repository ruvnet/ruvnet. Read docs/ecosystem-metrics-schedule.md and current repository instructions first. Extend the existing data/metrics.json, data/github-stats.json, data/registry-stats.json and dated data/snapshots structure; preserve existing schemas and unrelated content. Enumerate all public repos owned by ruvnet with complete pagination, including forks and archived repos, and report both all-owned and nonfork cohorts. Collect available stars, forks, subscribers, followers, issues, pull requests, releases, release asset download counts, and 14-day activity with explicit definitions and coverage. Refresh verified npm and crates.io package cohorts using existing collector logic where executable, keeping sources and metric-specific timestamps. Never claim code ran when it did not. Store dated source snapshots and update a running summary at data/ecosystem-running-totals.json plus a concise readable report at docs/ecosystem-metrics.md; link the report from the main README without replacing its design. Publish authorized metrics updates to the default branch, respecting branch protections and using a PR if required. Use fresh SHAs and avoid overwriting concurrent changes from the existing registry workflow. For clones and views, try an owner-authorized GitHub traffic interface available to ruOS. If unavailable, mark current traffic null with the exact blocker. Preserve prior traffic as historical. Store daily clone/view events per repository and UTC day; do not sum per-repository unique counts. Never expose credentials. A missing traffic day may be recorded as zero only when source repository metadata proves the repository did not exist on that date; otherwise store null and record the gap. Use a canonical daily ledger keyed by provider, stable repository/package ID, metric and UTC date. Upsert revised daily values; never add overlapping 14-day or 365-day windows. Exclude the incomplete current UTC day from cumulative daily totals. Record earliest date, missing days, collection coverage and backfills. Running clones/views/downloads must be labeled observed since the stated start date, not lifetime unless supported by provider data. Stars/forks/followers are current gauges with snapshot deltas, not cumulative sums. Keep npm, crates.io, release asset downloads, clones and views separate. Never sum unique cloners across repositories as unique people. Retain known release asset totals if assets disappear, with provenance and deletions recorded. Do not publish private repository names or data. Validate ledger uniqueness, replay idempotence, total reconciliation, pagination completeness and missing-data handling before committing. Unknown is null, never zero. Report commit or PR links, meaningful changes, coverage and actionable blockers. Do not send Slack/email messages.

Acceptance

After updating the JSON sources, run python3 scripts/render-visual-profile.py and include the regenerated profile SVGs in the same metrics commit. The dashboard headline uses download_milestone from data/ecosystem-running-totals.json; its npm growth chart retains the separate registry date window. Verify the milestone sum, remaining count, source dates and SVG rendering before publishing. If rendering cannot run, explicitly report the dashboard as stale rather than claiming it refreshed.

Replaying an identical snapshot must leave every running total unchanged. Every published total must include source, time window, cohort, coverage and missing-data status. No overlapping window totals may be added together.

For npm daily reporting, distinguish a completed UTC calendar day from a rolling 24-hour value. Report the latest confirmed daily total and date; keep recent all-zero dates provisional/unknown unless the provider confirms them. Use the existing package collector and preserve per-source timestamps.