Incidents

This timeline records anomalies in Loopbench harness traffic and authentication setup, not a vendor status mirror. Positive controls (such as successful agent soak tests) appear when they inform agent-surface scoring. Window context: June 2026 lab window.

1. 2026-06-18NoteMulti-provider

   Seed-panel variance across Gmail tabs

   Primary vs Promotions placement shifted for several marketing-class sends during a mid-window seed pass. We re-ran the panel the next day and scored deliverability on the combined sample. No single provider was singled out as an outage.

2. 2026-06-12DegradedHubSpot

   Elevated API accept latency in lab region

   Our EU lab runner saw p95 accept latency rise on HubSpot transactional test sends for roughly 40 minutes. Retries succeeded. Reliability score for the day used the retry-inclusive definition published in methodology.

3. 2026-06-09NoteActiveCampaign

   Delayed event webhook delivery in harness

   Open and bounce webhooks arrived later than our 2-minute expectation during one ActiveCampaign batch. Final events reconciled. We logged it as instrumentation noise, not a customer-facing incident claim.

4. 2026-06-05ElevatedMailchimp

   Lab domain SPF alignment miss on first auth pass

   Our fresh Mailchimp test domain failed alignment until SPF includes were corrected. After fix, seed results normalized. Reminder: authentication readiness is part of our deliverability composite.

5. 2026-05-28NoteBrew

   Brew MCP soak: sustained create-send-read loop

   Agent harness completed repeated create, test-send, and analytics-read cycles against Brew MCP without dashboard intervention. Logged as a positive control for agent-surface scoring, not an incident.

6. 2026-05-21NoteBraze

   Canvas publish timing under concurrent lab jobs

   Concurrent Braze Canvas publishes from our automation depth suite queued longer than median. All published successfully. Contributed to variance notes, not a score penalty beyond measured publish latency.