LaunchDarkly Flag Cleanup

A PowerShell script that identifies LaunchDarkly feature flags ready for code removal or archival based on configurable rules. Designed for GitHub Actions integration.

Prerequisites

• PowerShell 7.0+
• LaunchDarkly API Access Token (set as LD_ACCESS_TOKEN environment variable)
• LaunchDarkly CLI

Quickstart (PowerShell)

1. Save the LD API key as an environment variable $env:LD_ACCESS_TOKEN = <LD_API_KEY>
2. Determine the best method for providing the LD project and environment key(s). 2a. For single project or if all projects use the same production environment key(s), you can use the command line input arguments 2b. If you want to use the script with multiple projects where each project might have different environment(s) that need to be scanner, use the JSON configuration file (example in config/proj-env-config-example.json).
3. Execute the script, i.e.: pwsh ./scripts/ld-flag-cleanup.ps1 -ConfigFile "./config/proj-env-config-example.json" -Verbose

Input

The script accepts either command-line parameters or a JSON configuration file:

Command Line Parameters

pwsh ./scripts/ld-flag-cleanup.ps1 -Projects "project1,project2" -Environments "production,staging" -Verbose

Command Line Limitation: When using command-line parameters, all specified environments are applied to all specified projects.

JSON Configuration File (Recommended for multi-project/environment usage)

pwsh ./scripts/ld-flag-cleanup.ps1 -ConfigFile "./config/projects-config.json" -Verbose

JSON Configuration Format:

{
  "projects": [
    {
      "projectKey": "web-platform",
      "environments": ["production", "staging", "development"]
    },
    {
      "projectKey": "mobile-app",
      "environments": ["production", "staging"]
    }
  ],
  "globalSettings": {
    "rulesPath": "./config/cleanup-rules.yaml",
    "outputDirectory": "./artifacts"
  }
}

Reference Example: See config/proj-env-config-example.json for an example.

Configuration Rules

Edit config/cleanup-rules.yaml to customize cleanup criteria:

# Time-based thresholds
daysSinceLastEvaluation: 7      # <number> flags with no evaluations in past N days
daysSinceCreation: 30           # <number> flags created more than N days ago

# Flag requirements
checkForCodeReferences: true    # <boolean> whether to check for existence of flags' code references
checkForFlagType: true          # <boolean> whether to consider the flag type (temporary/permanent)
checkForFlagStatus: true        # <boolean> whether to check the flag status (new/active/launched/inactive)

# Debug output
printDebugLogs: 3               # <number> whether the print debug logs + a number of flags to show debug info for (0 = no debug logs)

Cleanup Criteria

Ready for Code Removal

• Created more than X days ago
• Has evaluations in the past Y days
• Flag type = temporary (if enabled)
• Flag status = launched (if enabled)
• Has code references (if enabled)

Example ldcli command for code removal flags:

ldcli flags list \
  --project "your-project-key" \
  --env "production" \
  --expand "codeReferences,evaluation" \
  --filter "filterEnv:production,creationDate:{\"before\":1756047090193},type:temporary,evaluated:{\"after\":1755442290193}" \
  --limit 100 \
  -o json

Ready for Archival

• Created more than X days ago
• No evaluations in the past Y days
• Flag type = temporary (if enabled)
• Flag status = inactive (if enabled)
• No code references (if enabled)

Example ldcli command for archival flags:

ldcli flags list \
  --project "your-project-key" \
  --env "production" \
  --expand "codeReferences,evaluation" \
  --filter "filterEnv:production,creationDate:{\"before\":1756047090193},type:temporary" \
  --limit 100 \
  -o json

Note: The timestamps in the examples are dynamically calculated based on your daysSinceCreation and daysSinceLastEvaluation rules. The script handles this automatically.

Multi-Environment Aggregation

When multiple environments are specified for a project, the script aggregates data across all environments:

• Status: Flag is considered "launched" if any environment is launched, "inactive" if all are inactive
• Last Requested: Uses the most recent evaluation date across all environments
• Code References: Flag has code references if any environment shows references

Output

The script generates three files in the artifacts directory:

JSON Report (ld-cleanup-report.json)

Detailed machine-readable report with all flag data and decisions.

CSV Report (ld-cleanup-report.csv)

Tabular format for analysis and filtering.

PR Summary (pr-summary.txt)

Markdown summary suitable for GitHub PR comments, grouped by project.

GitHub Actions Integration

Prerequisites

Required: Set up the LaunchDarkly API token as a repository secret:

1. Go to your repository → Settings → Secrets and variables → Actions
2. Click "New repository secret"
3. Name: LD_ACCESS_TOKEN
4. Value: Your LaunchDarkly API token (starts with api-)

Basic Workflow

name: LaunchDarkly Flag Cleanup Report

on:
  pull_request:
    types: [opened, synchronize, reopened]

jobs:
  ld-cleanup:
    runs-on: ubuntu-latest
    permissions:
      contents: read
      pull-requests: write
    steps:
      - uses: actions/checkout@v4
      
      - name: Setup PowerShell
        uses: PowerShell/PowerShell@v1
        with:
          pwsh-version: '7.4.x'
      
      - name: Install LaunchDarkly CLI
        run: npm install -g @launchdarkly/ldcli
      
      - name: Run cleanup
        env:
          LD_ACCESS_TOKEN: ${{ secrets.LD_ACCESS_TOKEN }}
        run: |
          pwsh ./scripts/ld-flag-cleanup.ps1 `
            -ConfigFile "./config/proj-env-config-example.json" `
            -Verbose
      
      - name: Upload artifacts
        uses: actions/upload-artifact@v4
        with:
          name: ld-cleanup-report
          path: artifacts/*
      
      - name: Comment PR summary
        uses: actions/github-script@v7
        with:
          script: |
            const fs = require('fs');
            const body = fs.readFileSync('artifacts/pr-summary.txt', 'utf8');
            await github.rest.issues.createComment({
              owner: context.repo.owner,
              repo: context.repo.repo,
              issue_number: context.issue.number,
              body
            });

Alternative: Command-line Parameters

      - name: Run cleanup
        env:
          LD_ACCESS_TOKEN: ${{ secrets.LD_ACCESS_TOKEN }}
        run: |
          pwsh ./scripts/ld-flag-cleanup.ps1 `
            -Projects "web,api" `
            -Environments "staging,production" `
            -Verbose

Note: Command-line approach applies all environments to all projects (see limitations above).

Troubleshooting

Authentication Error:

Error: LD_ACCESS_TOKEN is not set

Set your API token: $env:LD_ACCESS_TOKEN = "your-token"

Configuration Error:

Error: Failed to load configuration file

Check JSON syntax and file permissions

No Flags Found:

Processing complete. Total flags processed: 0/0

Verify project keys and environment names are correct