sst-stage-prune

sst-stage-prune supports identifying and pruning SST Stages.

When PR environments are left behind, they create drag:

• Cost creep from unused cloud resources.
• Operational noise from stale stages that look active but are not.
• Deployment failures from limit overruns.

This tool automates that hygiene by reconciling SST stages with currently open GitHub PRs and removing PR stages that no longer map to an active PR.

Example Issues

Here's a few errors that I have run into due to excessive stale PR Stages:

Cloud Front Function Limit

Having ~75 stale PR environments caused a deployment failure on an unrelated project in the same tenant due to too many Cloud Front Functions:

Site sst:aws:StaticSite → SiteCloudfrontFunctionRequest aws:cloudfront:Function
sdk-v2/provider2.go:572: sdk.helper_schema: creating CloudFront Function (thecaringplace-inf-PR-543-SiteCloudfrontFunctionRequest-uswnrdwz): operation error CloudFront: CreateFunction, https response error StatusCode: 400, RequestID: ecaa9052-7393-40c5-b752-80805190faaa, TooManyFunctions: : [email protected]

Requirements

• Node.js
• GitHub CLI (gh) installed and authenticated
• An SST project with a runnable SST script (default script name: sst)

Usage

Run from the root of your SST project:

sst-stage-prune [options]
sst-stage-prune cleanup-pr-stages [options]

You can also run locally in this repo with:

npm start -- [options]

Options

• --dry-run: List stale PR stages without removing them.
• --pkg-mgr <pkgMgr>: Package manager used to run SST (npm, pnpm, yarn). Default: npm.
• --sst-script <sstScript>: Script name used to invoke SST. Default: sst.
• --pr-stage-format <prStageFormat>: String template for PR stages, using {} as PR placeholder. Default: PR-{}.
• --pr-stage-pattern <prStagePattern>: Regex used to detect PR stages. Default: PR-\d+.

Example

sst-stage-prune cleanup-pr-stages --dry-run --pkg-mgr pnpm --sst-script sst