mridang/Runner Common Setup
Common first-step setup applied to every workflow — runner hardening via step-security/harden-runner, generic XDG cache over ~/.cache, and Docker image layer caching via an in-house bin-packing cache stage built on @actions/cache@v4 + the Docker Engine API. Pin this once at the top of every job.
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| harden-egress-policy | Passed to step-security/harden-runner. One of "audit" or "block". | no | audit |
| harden-allowed-endpoints | Newline-separated list of allowed endpoints, used when harden-egress-policy is "block". | no | "" |
| harden-disable-sudo | Passed to step-security/harden-runner. Disables sudo on the runner. | no | false |
| generic-cache | Enable a generic always-on cache over ~/.cache (XDG cache home). Tools that respect XDG_CACHE_HOME (pip, Go build cache, gh CLI, Helm, oras, others) get cross-run persistence for free. Set to "false" to skip. | no | true |
| docker-cache | Enable Docker image caching. Set to "false" to skip entirely. | no | true |
| docker-cache-key | Override the cache key. When empty, a sensible default is derived from runner.os and a slugified workflow name. The default deliberately excludes the job name so all jobs in a workflow share one cache; different workflows get separate caches. | no | "" |
| docker-cache-read-only | If "true", restore the Docker cache but do not save it on cache miss. Recommended for PR branches to avoid cache pollution. | no | false |
| docker-cache-max-fraction | Fraction of free disk to use as the docker-cache budget at post-step time. Float between 0 and 1. Default 0.6. The post-step bin-packs the delta of newly-pulled images (first-fit-decreasing) into this budget and skips any that don't fit, never failing the job on disk-full. | no | 0.6 |
| docker-cache-free-disk | If "true", remove ~20-25 GB of well-known pre-installed bloat from the runner (Android SDK, .NET, Haskell, CodeQL, PyPy, etc.) before the cache restore. Use this on workflows that pull many large images (e.g. polyglot test matrices) and would otherwise blow the runner's ~14 GB free-disk budget. Adds ~30 seconds. | no | false |
Outputs
no outputs