Retry Fatigue Cookbook

Retry fatigue reports help operators route unresolved handoff webhook failures by endpoint, owner, and escalation lane. Use these examples when the same target endpoint serves multiple review workloads and the incident handoff needs a narrower view than the full endpoint summary.

Focus One Owner

Use owner when a handoff should follow one workload owner:

curl "http://localhost:7701/v1/namespaces/my-app/review/handoff-webhooks/retry-fatigue?owner=alice"

This keeps the endpoint grouping intact, but only includes unresolved retry recommendations tied to Alice's saved handoff digests.

Focus One Escalation Lane

Use escalation_level when the handoff belongs to one escalation class:

curl "http://localhost:7701/v1/namespaces/my-app/review/handoff-webhooks/retry-fatigue?escalation_level=review_overdue"

Common lanes include review_overdue, source_trust_anomaly, and any custom escalation value saved in the review handoff digest.

Use A Preset Lane

Use preset when dashboards or handoff scripts should share a stable lane name:

curl "http://localhost:7701/v1/namespaces/my-app/review/handoff-webhooks/retry-fatigue?preset=review-overdue"

Built-in presets are review-overdue, source-trust-anomaly, and unassigned-review-overdue. JSON responses include the same preset metadata so clients can render the available lanes without hard-coding descriptions or assembling example filters. SDK and REST tests validate that payload against testdata/fixtures/retry-fatigue-presets.schema.json, keeping the dashboard-facing contract stable. Docs consumers can fetch the same contract from /schemas/retry-fatigue-presets.schema.json, or discover published contracts through /schemas/index.json.

Preset Reference

Dashboards can read the same preset metadata from retry fatigue JSON responses, but this table is the compact reference for operators and handoff tooling:

Preset	Expanded filters	Example REST query	Example GraphQL args	Intended handoff audience
review-overdue	escalation_level=review_overdue	preset=review-overdue	preset: "review-overdue"	Review owners or coordinators handling assigned or snoozed work that missed its review window
source-trust-anomaly	escalation_level=source_trust_anomaly	preset=source-trust-anomaly	preset: "source-trust-anomaly"	Source-quality reviewers investigating credibility drops, repeated refutations, or trust anomalies
unassigned-review-overdue	owner=unassigned, escalation_level=review_overdue	preset=unassigned-review-overdue	preset: "unassigned-review-overdue"	Queue triage or on-call reviewers who need to claim overdue work with no explicit owner

Explicit owner or escalation_level query parameters can narrow or override a preset-expanded value. Keep preset names stable in dashboards and scripts; change the expanded filters in one place when the handoff lane changes.

Focus An Owner And Lane

Combine both filters when the incident is specific to one owner inside one escalation class:

curl "http://localhost:7701/v1/namespaces/my-app/review/handoff-webhooks/retry-fatigue?owner=alice&escalation_level=review_overdue"

Use this for narrow handoffs where the endpoint is noisy but the actionable work belongs to a single lane.

Export Markdown For Handoffs

Add format=markdown to produce incident-ready notes:

curl "http://localhost:7701/v1/namespaces/my-app/review/handoff-webhooks/retry-fatigue?owner=alice&escalation_level=review_overdue&format=markdown"

The Markdown export includes endpoint totals, owner counts, escalation-level counts, status-family counts, and the latest failure details for the scoped fatigue view.

GraphQL Dashboard Query

Use the same filters in GraphQL dashboards:

query {
  reviewHandoffRetryFatigue(
    namespace: "my-app"
    preset: "review-overdue"
  ) {
    targetUrl
    candidates
    totalAttempts
    ready
    waiting
    owners {
      owner
      count
    }
    escalationLevels {
      escalationLevel
      count
    }
    lastStatusCode
    lastError
  }
}

Keep unfiltered endpoint fatigue dashboards for broad monitoring. Use filtered views for handoffs, paging lanes, and workload-specific follow-up.