feat(notify): Teams DM bell beside failure emails (T1)

[Jarrod S Doucette (jdoucett)]

Why

archive.py had zero Teams/webhook code — every [FAILED] / [STALE] / [EXCLUSION-ABORT] went out only as email (send_alert_email() → smtplib → smtp.purdue.edu), which gets tuned out. This is the "backup didn't happen" T1 tier of the settled notification-triage strategy and the highest-value gap: what actually stops must-see failure signals being missed.

What

A small stdlib-only notify_bell() that POSTs a Teams Adaptive Card beside the durable audit email. The email stays as the record; the bell is the attention signal. Card title is prefixed ACT NOW:.

Delivered as a 1:1 Teams DM, not a channel @mention (routing decision 2026-07-11; mirrors merged florasense-tools PR #106 --dm mode). The card carries no <at> / no msteams.entities: a DM already notifies, and a Power-Automate flow-bot chat post renders <at> entities as literal text. DM-vs-channel is decided by the Power Automate flow behind EARLY_TEAMS_WEBHOOK, not the payload — flipping to a DM is a Power-Automate web-UI step done separately from this code.

• notify_bell() / build_teams_card() — urllib.request only, imports inside the body (engine invariant chore(reclaim): hold AAPF landing-zone raw until promote loop verified #1). Card shape reused from florasense-tools/fortress/reship_watch_alert.py.
• Env-gated + DEFAULT-OFF on EARLY_TEAMS_WEBHOOK — silent no-op (no POST) until wired into the run env on Negishi. No mention env vars needed or read.
• Never breaks a run — only HTTP {200,202} = delivered; any error (absent/failed webhook, network) degrades to email-only with a logged reason and never raises. The webhook is a secret (sig) and is never printed/logged.
• Fired beside every LOCAL send_alert_email() failure site. The size-routing completion email ([SUCCESS]/[RESHIP-WATCH]) is deliberately not belled.
• Inline twin in send_to_fortress (self-contained, closes over write_log only, imports inside) for the exclusion companion-guard abort — the one alert that continues the run (so it never reaches the local Step-2 wrapper) and can fire on a compute node under --globus. All other verification [FAILED]s raise → the Step-2 wrapper alerts and bells.

Tests

tests/test_notify_bell.py — 14 tests (POST mocked, os.environ fully controlled):

• env-gated no-op (no webhook → no POST), DM-clean payload (no <at>/no msteams.entities, legacy mention env vars ignored), ACT NOW: prefix + idempotency, HTTP 200/202 success, never-raises on HTTPError / URLError / generic exception + non-2xx, and webhook-never-printed secret safety.

python3 -m unittest discover -s tests → 289 passed.

Deploy note (default-off)

Does nothing until EARLY_TEAMS_WEBHOOK is exported into the run environment on Negishi (archive_local.sbatch / cron), and the Power Automate flow behind it is set to post a DM. Until then notify_bell() is a silent no-op and email behavior is unchanged.

Scope guards

• Does not touch the reship-watch alert (already correct).
• No --dm / --task modes — triage design keeps everything on the one alert surface.

🤖 Generated with Claude Code