All jobs

Add OpenSSF Scorecard to a Repository with GitHub Actions

Yes. An agent adds the OpenSSF Scorecard workflow, publishes its results to code scanning, pins every action by commit so the checks pass, and opens a pull request.

Run this on RushOpens the console with this prompt ready to send, in a Project that has your repository.

What the agent does

The agent reads every workflow before it adds one, because Scorecard grades the repository as a whole and most of its findings are about files that already exist. In a repository started from actions/typescript-action there are five workflows: ci.yml runs format, lint and Jest on the Node version in .node-version, which is 24.4.0; check-dist.yml rebuilds dist/ with Rollup and fails on a diff; codeql-analysis.yml runs github/codeql-action on src with a weekly cron; linter.yml runs super-linter's slim image with .checkov.yml and a VALIDATE_GITHUB_ACTIONS_ZIZMOR: false switch; and licensed.yml checks cached licenses on workflow_dispatch. Each sets top-level permissions with contents: read, and .github/dependabot.yml already updates github-actions and npm weekly in groups. CODEOWNERS and LICENSE exist; there is no SECURITY.md and no Scorecard workflow.

With that inventory the agent knows which Scorecard checks the repository will pass as it is, which it can fix in the diff, and which depend on settings it cannot reach. Dependency-Update-Tool, SAST, License and the per-workflow Token-Permissions are in place. Pinned-Dependencies is not: every uses: line references a tag such as actions/checkout@v6, actions/setup-node@v6, github/codeql-action/init@v4, ruby/setup-ruby@v1, super-linter/super-linter/slim@v8 and actions/upload-artifact@v7, and Scorecard wants a commit SHA. The agent resolves each tag to the commit it points at with the GitHub API and rewrites it as @<sha> # vX.Y.Z, the comment form Dependabot keeps updated, so the groups in dependabot.yml carry on working. Security-Policy it fixes by adding a short SECURITY.md that says where to report a vulnerability, with a placeholder contact the reviewer fills in. Branch-Protection, Code-Review and Signed-Releases live in repository settings and release practice, so the pull request explains them rather than pretending to fix them.

The new .github/workflows/scorecard.yml runs ossf/scorecard-action, itself pinned by SHA, on pushes to the default branch, on branch_protection_rule events and on a weekly schedule, with permissions limited to security-events: write for the SARIF upload, id-token: write so publish_results: true can sign the result, and contents: read and actions: read for the checks that read workflows and runs. The SARIF goes to code scanning through github/codeql-action/upload-sarif, the same action family the CodeQL workflow already uses, so findings appear next to CodeQL's in the Security tab. The agent also notes the actionlint.yml rule that ignores invalid runner name "node24", since action.yml declares runs.using: node24, and leaves it in place.

To check its own work, the agent runs actionlint over all six workflows in its sandbox, runs zizmor on them because super-linter has that check switched off, and, where the repository is public, runs the Scorecard CLI against the branch to see which checks moved. It cannot see the published result for the pull request branch, because publishing happens only on the default branch, so the description says which checks it expects to change and why.

Start it with one API call

POST https://api.phoenix.so/o/{org}/sessions with this body:

{
  "agent": "claude",
  "input": "Add OpenSSF Scorecard to this repository. Read every workflow under .github/workflows, .github/dependabot.yml, action.yml, actionlint.yml and CODEOWNERS first. Add a scorecard workflow that runs on pushes to the default branch, branch protection changes and a weekly schedule, uploads the SARIF to code scanning and publishes results, with the smallest permissions that allows. Then fix what Scorecard would flag in the files themselves: pin every action in every workflow to a commit SHA with a version comment that Dependabot can keep updated, and add a SECURITY.md. Run actionlint and zizmor on the workflows. Do not change repository settings; list the checks that need them in the pull request, and open it.",
  "resources": [
    {
      "url": "https://github.com/your-org/your-app"
    }
  ]
}
  1. Get an API key. In the console, open Settings, then API keys, and create a key for your org. Send it in the Authorization header as a Bearer token.
  2. Send the request. POST the body in request.json to https://api.phoenix.so/o/{org}/sessions, with your repository as the resources URL. The run starts in its own cloud sandbox.
  3. Read what came back. Follow the run on its live stream or wait for the run.completed webhook, then read the result and its trajectory.

What comes back

What comes back is a pull request on its own branch adding .github/workflows/scorecard.yml and SECURITY.md, and touching each existing workflow only on its uses: lines, where tags become commit SHAs with version comments. The steps, permissions and triggers of the existing workflows are unchanged, so CI, CodeQL, the linter and the dist check keep doing what they did.

The description has a table of the Scorecard checks: the ones the repository passed already, the ones this diff addresses and the ones that need a setting, with the setting named for each, such as branch protection on the default branch requiring reviews. It quotes the actionlint and zizmor output and, when it could run the Scorecard CLI, the before and after for the checks it changed. It also adds the Scorecard badge to the README in a form that renders after the first run on the default branch.

Before merging, a reviewer confirms each SHA matches the version in its comment, decides whether id-token: write is acceptable in that repository, and fills in the contact in SECURITY.md; after merging, the first scheduled run publishes the score and the badge starts showing it.

Run it on every new repository in your organization

GitHub signs its webhooks and cannot add the Authorization header a Rush webhook trigger requires, so a small handler on your server receives the organization's repository event, checks its signature and sends this body to POST /o/{org}/sessions with the new repository in resources. It sends a request_id that is a UUID v5 of the delivery id, so GitHub's retries return the run already started instead of starting another.

With a webhook_url on the key, Rush POSTs run.completed when the turn ends; the handler reads the pull request from the session and posts it where the repository's owners will see it. Every repository created in the organization then starts with Scorecard and pinned actions from its first week, with the transcript showing what was checked.

The use case this job follows, with its automation and trigger: Hand off a long task.

Why run it on Rush?

Scorecard is one workflow to add and a dozen checks to understand, and the understanding is the work: which findings are about files, which about settings, and how to pin actions without breaking the Dependabot groups that keep them current. The agent reads all five workflows and the Dependabot config first, so the pull request fixes what a diff can fix and says plainly what it cannot.

On Rush the pinning, the linting and the local Scorecard run happen in the agent's own sandbox with the real tools, and the transcript keeps every command and its output next to the diff. An organization with many repositories can hand each one to its own run, each with its own sandbox and its own pull request, rather than copying one workflow around by hand.

Frequently asked questions

Why does the workflow need id-token: write?

Publishing results to the OpenSSF API signs them with the workflow's OIDC token, so publish_results: true needs id-token: write. Without publishing, the SARIF upload to code scanning still works and only security-events: write is needed; the pull request says which mode it set and how to switch.

Will Dependabot still update the pinned actions?

Yes. Dependabot updates SHA pins when the line carries a version comment, and the agent writes every pin as @<sha> # vX.Y.Z. The existing weekly github-actions entry with its minor and patch group applies to the new workflow as well.

Which checks cannot be fixed from the repository?

Branch-Protection, Code-Review and Signed-Releases depend on settings and release practice, and Maintained on activity. The pull request lists each with the setting to change, such as requiring a review before merging to the default branch, instead of adding files that would not move the score.

Does the score show on the pull request?

No. The Scorecard action publishes only from the default branch, so the pull request shows the workflow and the pinned actions, and the first run after merge produces the score and the badge. The agent's local run of the Scorecard CLI, where it could do one, is quoted in the description as a preview.

More GitHub Actions jobs

The same job on other stacks

More jobs like this: Hand off a long task

Keep reading