Scorecards as code

Once Scorecards are incorporated into your regular engineering workflows, they become production-grade tools. Scorecards set the benchmarks for how your organization defines “good,” which impacts everything from security standards to on-call rotations. Because Scorecards are crucial to achieving and maintaining standards, you likely want to employ some kind of controlled access and version controlling.

With Scorecards as code, you can treat your Scorecard like any other production asset. This feature lets you manage Scorecards through GitOps, just like you’d manage the entities in your catalogs. Every Scorecard has its own YAML file (you can learn more about Scorecard YAML formatting below under GitOps Scorecard best practices). When GitOps is enabled, changes made to Scorecards will appear in GitOps logs, just like when changes are made to other entities.

Setting up Scorecards as code

To migrate to Scorecards as code:

1. Navigate to Settings > GitOps then click the Scorecards tab.

2. Toggle off the Enable Scorecard UI editor option.

In the GitOps settings, click the Scorecard tab.

When you disable the Scorecard UI editor, it only affects Scorecards that are backed by a YAML file. You will still be able to create Scorecards via the Cortex UI, and you can still edit a Scorecard via the UI if it is not being managed through GitOps.

Converting an existing Scorecard to code

To convert an existing Scorecard into code:

1. Navigate to the Scorecard's details page.

2. Click the vertical ellipsis (⋮), then click Export CSV.

This will download a YAML version of your Scorecard.

Managing Scorecards via GitOps is designed to reduce as much friction as possible, so you can continue to make the most of this flexible tool.

GitOps Scorecard best practices

In general, the best place to put Scorecards is in their own repository, separate from catalog entities, at the repository's root directory within .cortex/scorecards. Note that it is not recommended to put Scorecard definitions in a service repository as Scorecards are not meant to be 1:1 with catalog entitites. For example, a simple repository might have the structure:

.
├── .cortex
│    └── scorecards
│        ├── dora.yml
│        └── performance.yml
└── src
    └── index.js
    └── ...

Any file found within the .cortex/scorecards directory will be automatically picked up and parsed as a Scorecard.

Example Scorecard YAML file

The dora-metrics-scorecard.yaml descriptor file might look something like this:

name: DORA Metrics
tag: dora-metrics
description: >-
  [DORA metrics](https://www.cortex.io/post/understanding-dora-metrics) are used by DevOps teams to measure their performance.

  The 4 key metrics are Lead Time for Changes, Deployment Frequency, Mean Time to Recovery, and Change Failure Rate.
draft: false
notifications:
  enabled: true
  scoreDropNotificationsEnabled: false
exemptions:
  enabled: true
  autoApprove: false
ladder:
  levels:
    - name: Bronze
      rank: 1
      description: Pretty good
      color: "#c38b5f"
    - name: Silver
      rank: 2
      description: Very good
      color: "#8c9298"
    - name: Gold
      rank: 3
      description: Excellent
      color: "#cda400"
rules:
  - title: Ratio of rollbacks to deploys in the last 7 days
    expression: >+
      (deploys(lookback=duration("P7D"),types=["ROLLBACK"]).length /
      deploys(lookback=duration("P7D"),types=["DEPLOY"]).length) > 0.05
    identifier: 3c42fa96-b422-30a4-b75a-f8b1cc233408
    description: Change Failure Rate
    weight: 25
    failureMessage: Less than 95% of deployments in the last 7 days were successful
    level: Gold
  - title: Incident was ack'ed within 5 minutes
    expression: oncall.analysis(lookback = duration("P7D")).meanSecondsToFirstAck <= 300
    identifier: 8713f2c0-f161-3688-9f99-bcfaab476b63
    description: MTTA (Mean time to acknowledge)
    weight: 25
    failureMessage: Incidents in the last 7 days were not ack'ed
    level: Silver
  - title: Last commit was within 24 hours
    expression: git.lastCommit().freshness = 7
    identifier: a16b7eeb-545b-359e-81a7-3946baacdd4b
    description: Deployment Frequency
    weight: 25
    failureMessage: No deployments in the last 7 days
filter:
  kind: GENERIC
  types:
    include:
      - service
  groups:
    include:
      - production
evaluation:
  window: 4

Scorecard YAML reference

Objects

Scorecard: 

name	description
name	The human-readable name of the Scorecard
tag	A unique slug for the Scorecard consisting of only alphanumeric characters and dashes
description	A human-readable description of the Scorecard
draft	Whether or not the Scorecard is a draft
notifications	Notifications settings for the Scorecard
ladder	The ladder to apply to the rules
rules	A list of rules that are evaluated each time the Scorecard is evaluated
filter	Enables the ability to exclude entities from being evaluated by this Scorecard
evaluation	Enables the ability to change the evaluation window for this Scorecard

Notifications: 

name	description
enabled	Whether or not to include the Scorecard in notifications
scoreDropNotificationsEnabled	Whether to notify about entities' scores that dropped after each evaluation of this Scorecard

Exemptions: 

name	description
enabled	Whether or not rule exemptions are enabled for Scorecard
autoApprove	Whether or not rule exemptions are auto approved for Scorecard

Ladder: 

name	description
levels	The levels of the ladder

Level: 

name	description
name	The human-readable name of the level
rank	The rank of the Level within the ladder. Higher rank is better.
description	A human-readable description of the level
color	The hex color of the badge that is displayed with the level

Rule: 

name	description
title	The human-readable name of the Rule
expression	The CQL expression to evaluate; must evaluate to a boolean
identifier	Identifier of the rule, unique within Scorecard scope.
description	A human-readable description of the Rule
weight	The number of points this Rule provides when successful
failureMessage	A human-readable message that will be presented when the Rule is failing
level	The name of the level this rule is associated with; can be null even when a ladder is present
filter	Enables the ability to exclude entities from being evaluated for this rule
effectiveFrom	Date when the rule starts being evaluated (e.g. 2024-01-01T00:00:00Z)

Filter: 

Note: One of types, groups, query must be present for it to be considered a valid filter

name	description
kind	The kind of filter to create. Currently only supports "GENERIC"
types	Types filter (to include / exclude specific types)
groups	Groups filter (to include / exclude specific groups)
query	A CQL query; only entities matching this query will be evaluated by the Scorecard

TypesFilter: 

Note: Only one of include/exclude can be specified at a time

name	description
include	List of types to include in set of entities
exclude	List of types to exclude in the set of entities

GroupsFilter: 

name	description
include	List of groups to include in set of entities
exclude	List of groups to exclude in the set of entities

Evaluation: 

name

description

window

By default, Scorecards are evaluated every 4 hours. That default can be changed under Settings > Scorecards > Default evaluation window. If you would like to evaluate Scorecards less frequently, you can override the evaluation window. This can help if you're encountering problems with rate limits. Note that Scorecards cannot be evaluated more than once per minimum set in Settings > Scorecards > Minimum evaluation window.