fix: include community collections in collectionsSlugs (IN-1195)

[gaspergrom]

Problem

Community (non-LF) collection detail pages show Contributors: 0 and Avg. Health: — even when the collection has projects (e.g. agentic-ai-momentum-watchlist shows 88 projects but 0 contributors). Reported from a review of the Collections feature.

Root cause is in the Tinybird layer, not the frontend. The collection_insights_aggregate pipe (which powers the header metrics) resolves collection membership via insights_projects_populated_ds.collectionsSlugs. That array is built in insights_projects_populated_copy.pipe, whose collectionsSlugs node filtered with AND isNull(c.ssoUserId) — ssoUserId is set for community/user-created collections and null only for curated (LF) ones. So every community collection was excluded from collectionsSlugs, and the aggregate pipe matched 0 projects for them → uniqueContributorCount: 0, avgHealthScore: nan. The Postgres-derived project count still worked, which is why the count showed but the other two metrics didn't.

Verified against prod Tinybird:

• cloud-native-computing-foundation-(cncf) (curated) → 127 / 306251 / 72 ✅
• agentic-ai-momentum-watchlist (community) → 0 / 0 / nan ❌
• arrayJoin(collectionsSlugs) contains no community slugs (only curated).

Fix

Drop the AND isNull(c.ssoUserId) filter from the collectionsSlugs node so community collection memberships are populated alongside curated ones.

Blast radius

Other consumers of collectionsSlugs (collection_insights_aggregate, segments_filtered_by_collection, leaderboards_*, insightsProjects_filtered) only apply a collectionsSlugs filter when an explicit collectionSlug param is passed — so no un-scoped listing changes. The only new capability is that passing a community collectionSlug now returns results, which is the intended behavior.

insights_projects_populated_copy is a COPY pipe on a */15 * * * * schedule, so the 43 existing community collections will backfill correct metrics on the next run after deploy — no manual backfill needed.

Validation

• tb check passes
• tb fmt applied
• No schema change; datasource collectionsSlugs column unchanged (only more rows populated)

Related

Frontend collection-widget wiring (Contributors/Popularity/Development tabs) is tracked separately in insights PR #2013 (IN-1191). This PR is the upstream data half and is independent of it.

https://claude.ai/code/session_014Fc86j8FLeWFAs3fDCQs38

[cursor]

PR Summary

Low Risk
Single WHERE-clause change in a scheduled COPY pipe; scoped to collection slug filtering with no schema changes.

Overview
Fixes community collection header metrics (Contributors: 0, Avg. Health: —) by changing how collectionsSlugs is built in insights_projects_populated_copy.pipe.

The collectionsSlugs node no longer filters out collections where ssoUserId is set, so community memberships are written into insights_projects_populated_ds alongside curated ones. That lets collection_insights_aggregate resolve projects when a community collectionSlug is passed.

Downstream pipes only filter on collectionsSlugs when an explicit collectionSlug is provided, so un-scoped listings should be unchanged. The COPY pipe’s 15-minute schedule is expected to backfill existing community collections after deploy.

^{Reviewed by Cursor Bugbot for commit 49f1949. Bugbot is set up for automated code reviews on this repo. Configure here.}

[Copilot]

Pull request overview

Updates Tinybird collection membership so public community collections receive contributor and health metrics.

Changes:

• Removes the curated-only collection filter.
• Documents downstream impact.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.