Hosted application
Tenhaw Delivery Health
Review one service's delivery health from a supplied aggregate metrics report. Keep reporting windows, sample sizes and unavailable observations visible. The hosted app accepts aggregate metrics only; it does not collect repository events or rank unrelated teams.
Delivery and engineering leaders reviewing a defined service and reporting period.
Source preview. The package has not been published as a ChatGPT directory listing. Use the source and verification records to review the implementation.
Start with your decision
- Review this aggregate report for one delivery service.
- Explain which delivery measures lack production evidence.
- Turn these aggregate delivery observations into improvement questions.
How it works
- Ask for the service and reporting window and a supplied aggregate DeliveryReport. The report has period, metrics and caveats. Use its metric values, units, sampleSize, period, dataSource and caveats. A narrative estimate is not a substitute for deployment evidence.
- Call score_delivery_health with {"metrics":<aggregate report>}. Preserve null values and sample sizes. DORA measures need linked production, commit and incident evidence; tags, merges and deployment-job failures are not interchangeable with observed production outcomes.
- Call show_delivery_health with the same aggregate input. Explain the observed measures and practical evidence improvements. Do not invent an overall team league table or industry performance band.
- The hosted app has no event collector. For raw GitHub/Jira events, explain that collection and aggregation must happen in the user's local environment, then request only the resulting aggregate report. Never request credentials or transmit raw repository events to the hosted endpoint.
Tools and connection
The scoped connection offers only these operations:
score_delivery_healthshow_delivery_health
https://delivery-health.apps.tenhaw.com/mcp/apps/delivery-healthNo account or Tenhaw API key is required by these tools. Calculation inputs are processed transiently. The results view lets you revise assumptions and export a report.
Supply an aggregate delivery report. Source-system collection and raw repository events stay in your own environment.