---
title: "Get Dashboard"
method: GET
path: "/regression/dashboard"
---

# Get Dashboard

`GET /regression/dashboard`

Aggregate stats for the Regressions tab header.

## Query parameters

- `trend_window_days` integer

## Response `200`

Successful Response

- RegressionDashboardResponse — Aggregate response for the Regressions tab header.
  - `golden_set_size` integer, required
  - `modes` RegressionModeDTO[], required
    - `mode` 'smoke' | 'default' | 'comprehensive', required — Named regression modes. Distinct from ``ThinkingMode`` (the agent's reasoning budget): a regression mode bundles *everything* — incident count, reruns, thinking budget, and verdict thresholds — into one choice.
    - `label` string, required
    - `description` string, required
    - `thinking_mode` string, required
    - `runs_per_incident` integer, required
    - `max_incidents` integer, nullable
    - `judge_rounds` integer, required
    - `max_parallel_pods` integer, required
    - `apply_pairwise_verdict` boolean, required
    - `gt_drop_warn` number, required
    - `gt_drop_fail` number, required
    - `per_incident_gt_drop_warn` number, required
    - `per_incident_gt_drop_fail` number, required
    - `max_incident_regressions_fail` integer, required
    - `pairwise_loss_rate_warn` number, required
    - `pairwise_loss_rate_fail` number, required
    - `duration_increase_warn_pct` number, required
  - `default_mode` 'smoke' | 'default' | 'comprehensive', required — Named regression modes. Distinct from ``ThinkingMode`` (the agent's reasoning budget): a regression mode bundles *everything* — incident count, reruns, thinking budget, and verdict thresholds — into one choice.
  - `last_30_days` RegressionSummaryCounts, required — Counts of regression verdicts across a time window.
    - `passed` integer
    - `warn` integer
    - `fail` integer
    - `running` integer
    - `failed_to_run` integer
    - `unknown` integer
    - `total` integer
  - `trend` RegressionTrendPoint[], required
    - `created_at` string, date-time, required
    - `verdict` 'passed' | 'warn' | 'fail' — Regression test verdict, mirrors ``regression_verdict_enum``. A verdict is a regression-testing concept: pass / warn / fail against the curated golden set. Manual A/B runs (``origin='manual'``) never have a verdict — it lives on ``ab_tests`` only because that's where regression runs are persisted. Note the enum value for PASS is ``"passed"`` (not ``"pass"``) to avoid colliding with the Python keyword and to match the Postgres enum. Clients filtering on this field must use ``"passed"``.
    - `name` string, required
    - `branch_b` string, required
    - `gt_score_a` number, nullable
    - `gt_score_b` number, nullable
  - `verdict_reasons_sample` object[] — Latest verdict_reasons payload, for quick diagnostics.

## Other responses

- `422` — Validation Error

---

[API](https://skmtc.net/traversal/apis/fastapi.md) · [All operations](https://skmtc.net/traversal/apis/fastapi/llms.txt) · [OpenAPI document](https://skmtc-service-staging.skmtc.workers.dev/v1/apis/traversal/fastapi/revisions/2134ebffd1ef/schema)