Daily Slack digest of WYRE product adoption for Conduit (conduit.wyre.ai).
Pulls /api/admin/metrics:
- Active orgs (30-day rolling)
- Total tool calls and per-org breakdown
- Top vendors by tool calls
- Plan distribution
- New org signups
Runs every day at 14:05 UTC (five minutes after stars-watcher so the two
digests don't land on top of each other). The two watchers are deliberately
complementary: stars-watcher tracks external reach (stars, registry
coverage, PulseMCP traffic); this one tracks internal usage.
Renamed from
gateway-adoption-watcheron 2026-08-20 when Conduit coverage was added; GitHub redirects the old URL. The legacy MCP gateway (mcp.wyre.ai) has since been retired. This job does not fetch it, and there is noGATEWAY_BASE/GATEWAY_ADMIN_TOKENsetting.
- Delivery uses the shared WYRE Notifier Slack app (WYRE AI workspace,
manifest in
wyre-technology/.github→slack-app/notifier/) via the org-levelSLACK_NOTIFIER_BOT_TOKENActions secret — nothing to set per repo. The target channel isSLACK_CHANNEL_IDin.github/workflows/daily.yml(currentlyC0BSHBQQBQ8, #product-notifications). - Add the Conduit admin token (pipe straight from Azure — never paste values):
# Conduit — conduit-prod Container App secret az containerapp secret show -n conduit-prod-gateway -g rg-conduit-prod \ --secret-name admin-api-key --query value -o tsv \ | gh secret set CONDUIT_ADMIN_TOKEN --repo wyre-technology/adoption-watcher - (Optional) Override the Conduit base URL — defaults to
https://conduit.wyre.ai:gh variable set CONDUIT_BASE --body "…" --repo wyre-technology/adoption-watcher - Trigger manually to verify formatting:
gh workflow run daily.yml --repo wyre-technology/adoption-watcher
script/report.py is plain stdlib Python. It hits Conduit's metrics
endpoint with its bearer token, formats a Slack Block Kit message, and
posts it via the WYRE Notifier bot. It saves the response to
state/snapshot.json (schema: 2, keyed by product) and diffs against the
previous run for "since last run" deltas. A leftover gateway key, or an
older flat pre-schema gateway payload, is ignored and not written back.
If the fetch fails, the section is replaced by a warning block and the run exits non-zero so the Actions run shows red.
The metrics endpoint returns 30-day rolling counts. When it grows a
?window=24h query parameter, swap that in for a true day-over-day delta —
same digest shape.