Skip to content
Sextant

Outputs

The action sets three outputs once sextant grade --pr finishes. Use them to chain follow-up steps — gating deploys, notifying Slack, and similar.

OutputDescription
verdictAPPROVE or REQUEST_CHANGES.
new-errorsCount of new error-severity findings introduced by the PR.
new-warningsCount of new warn-severity findings introduced by the PR.

The action’s exit status is the canonical signal — outputs let downstream steps inspect the result without re-running the grade.

Example: Slack notification on regression

Section titled “Example: Slack notification on regression”
- id: sextant
  uses: kylebastien/sextant-mcp/action@v0.1.0


- if: steps.sextant.outputs.verdict == 'REQUEST_CHANGES'
  uses: slackapi/slack-github-action@v1
  with:
    channel-id: code-review
    slack-message: |
      :warning: PR ${{ github.event.pull_request.number }} introduces
      ${{ steps.sextant.outputs.new-errors }} new error
      and ${{ steps.sextant.outputs.new-warnings }} new warn findings.
      ${{ github.event.pull_request.html_url }}

Example: gating a deploy

Section titled “Example: gating a deploy”
- id: sextant
  uses: kylebastien/sextant-mcp/action@v0.1.0
  with:
    fail-on: never        # let the deploy step decide


- name: Deploy
  if: steps.sextant.outputs.verdict == 'APPROVE'
  run: ./scripts/deploy.sh

fail-on: never keeps the action exit code at 0 even on regression, so the workflow doesn’t short-circuit; the deploy step then guards on the output explicitly.

Example: counts in a job summary

Section titled “Example: counts in a job summary”
- id: sextant
  uses: kylebastien/sextant-mcp/action@v0.1.0


- run: |
    echo "## Sextant" >> "$GITHUB_STEP_SUMMARY"
    echo "Verdict: ${{ steps.sextant.outputs.verdict }}" >> "$GITHUB_STEP_SUMMARY"
    echo "New errors: ${{ steps.sextant.outputs.new-errors }}" >> "$GITHUB_STEP_SUMMARY"
    echo "New warnings: ${{ steps.sextant.outputs.new-warnings }}" >> "$GITHUB_STEP_SUMMARY"

See also

Section titled “See also”