---
title: "List Agent Score history"
method: GET
path: "/v1/projects/{projectSlug}/agent-score/history"
tags: ["Agent Score"]
---

# List Agent Score history

`GET /v1/projects/{projectSlug}/agent-score/history`

Returns the project's published Agent Scores in a date range, oldest first. Days the project did not publish are absent from the list.

## Path parameters

- `projectSlug` string, required — Project slug (human-readable identifier)

## Query parameters

- `from` string, date — Inclusive start date as `YYYY-MM-DD`. Defaults to 90 days before `to`.
- `to` string, date — Inclusive end date as `YYYY-MM-DD`. Defaults to today.

## Response `200`

Published Agent Scores in the range

- AgentScoreHistory
  - `snapshots` AgentScoreSnapshot[], required — Published scores in the range, oldest first. Days without a published score are absent, not zero.
    - `date` string, required — UTC date the score was published for, as `YYYY-MM-DD`.
    - `score` number, required — Agent Score from 0 to 100: the weighted mean of the five dimensions for this window.
    - `interval` AgentScoreInterval, required
      - `lower` number, required — Lower bound of the 95% interval, on the same 0 to 100 scale.
      - `upper` number, required — Upper bound of the 95% interval, on the same 0 to 100 scale.
    - `dimensions` AgentScoreDimensions, required
      - `outcome` AgentScoreDimension, required — Share of comparable sessions expected to accomplish the user's task.
        - `score` number, required — Dimension score from 0 to 100. Read it with the dimension's own meaning.
        - `interval` AgentScoreInterval, required
          - `lower` number, required — Lower bound of the 95% interval, on the same 0 to 100 scale.
          - `upper` number, required — Upper bound of the 95% interval, on the same 0 to 100 scale.
      - `reliability` object, required — Chance the agent completes a reference run of 20 sessions without a terminal operational failure.
        - `score` number, required — Dimension score from 0 to 100. Read it with the dimension's own meaning.
        - `interval` AgentScoreInterval, required
          - `lower` number, required — Lower bound of the 95% interval, on the same 0 to 100 scale.
          - `upper` number, required — Upper bound of the 95% interval, on the same 0 to 100 scale.
      - `cost` object, required — Health of the agent's use of paid and token-bearing resources.
        - `score` number, required — Dimension score from 0 to 100. Read it with the dimension's own meaning.
        - `interval` AgentScoreInterval, required
          - `lower` number, required — Lower bound of the 95% interval, on the same 0 to 100 scale.
          - `upper` number, required — Upper bound of the 95% interval, on the same 0 to 100 scale.
      - `speed` object, required — Share of user-visible critical-path time that was necessary.
        - `score` number, required — Dimension score from 0 to 100. Read it with the dimension's own meaning.
        - `interval` AgentScoreInterval, required
          - `lower` number, required — Lower bound of the 95% interval, on the same 0 to 100 scale.
          - `upper` number, required — Upper bound of the 95% interval, on the same 0 to 100 scale.
      - `safety` object, required — Chance a reference run of 100 sessions contains no confirmed agent-caused harm.
        - `score` number, required — Dimension score from 0 to 100. Read it with the dimension's own meaning.
        - `interval` AgentScoreInterval, required
          - `lower` number, required — Lower bound of the 95% interval, on the same 0 to 100 scale.
          - `upper` number, required — Upper bound of the 95% interval, on the same 0 to 100 scale.
    - `scoringVersion` string, required — Version of the formulas, prompts and frozen references that produced this score. Scores from different versions are not directly comparable.
    - `windowDays` integer, required — Length in days of the rolling window this score covers.
    - `eligibleSessionCount` integer, required — Production sessions in the window the score was computed over.
    - `policyCap` number, nullable, required — Ceiling a safety policy rule applied to the score, or `null` when no rule applied.

## Other responses

- `400` — Validation error
- `401` — Unauthorized
- `404` — Not found

## Changes

- **2026-09-12** `c726d219e721` — 1 info
  - endpoint added

[Change history](https://skmtc.dev/latitude-dev/apis/latitude/changes/v1/projects/:projectSlug/agent-score/history/get.md)

---

[API](https://skmtc.dev/latitude-dev/apis/latitude.md) · [All operations](https://skmtc.dev/latitude-dev/apis/latitude/llms.txt) · [OpenAPI document](https://skmtc.dev/latitude-dev/apis/latitude/revisions/44d01c7b9918?raw)