drt docs design lab
icon strategy
Design lab

Connector icons — A/B/C — decided: B (2026-07-03)

Same node cards, three icon strategies. Judge at a glance: which tells you "that's BigQuery" fastest, and which stays classy at 20+ connectors?

A — generic glyphs

bq_prod
BIGQUERY
pg_replica
POSTGRES
#alerts
SLACK
contacts
HUBSPOT
exports
S3

Pro: zero maintenance, perfectly consistent, no trademark questions. Con: every DB looks identical — you must read the type label.

B — brand-color initial badges ✅ adopted

BQ
bq_prod
BIGQUERY
PG
pg_replica
POSTGRES
SL
#alerts
SLACK
HS
contacts
HUBSPOT
S3
exports
S3

Pro: instant recognition via brand color, scales automatically to ANY connector (initials derived from the type string), no logo files, no trademark art. Con: colors are "borrowed" branding; two-letter codes need a collision rule; palette needs a contrast pass per badge (dark text on light brands, as HS/S3 above). Adopted as the system — the curated map + neutral fallback lives in drt/docs/html.py (_BADGES).

C — vendored official logos (dbt docs / dagster style)

logo
svg
bq_prod
BIGQUERY
logo
svg
pg_replica
POSTGRES
logo
svg
#alerts
SLACK

(placeholders — not drawing trademarks by hand) Pro: best recognition, the "rich" look dbt/dagster have. Con: vendor ~20 SVGs + licence/trademark review per logo, ongoing maintenance as connectors land, monochrome/dark variants needed, and a fallback badge is still required for unknown plugins — so C in practice = B + logo overrides. Possible later enhancement on top of B.