action.yml

name: Agent Skill Doctor
description: Statically check an Agent Skill without executing its scripts.
author: Agent Skill Doctor contributors

branding:
  icon: shield
  color: blue

inputs:
  path:
    description: Path to the Skill directory to check.
    required: false
    default: .
  format:
    description: Report format (text, json, or sarif).
    required: false
    default: text
  output:
    description: Optional report file. Leave empty to write to standard output.
    required: false
    default: ""
  fail-on:
    description: Minimum severity that fails the step (error, warning, critical, or never).
    required: false
    default: error

runs:
  using: composite
  steps:
    - name: Set up Python
      uses: actions/setup-python@v5
      with:
        python-version: "3.12"

    - name: Install Agent Skill Doctor
      shell: bash
      run: python -m pip install --disable-pip-version-check "$GITHUB_ACTION_PATH"

    - name: Check Skill
      shell: bash
      env:
        SKILL_DOCTOR_PATH: ${{ inputs.path }}
        SKILL_DOCTOR_FORMAT: ${{ inputs.format }}
        SKILL_DOCTOR_OUTPUT: ${{ inputs.output }}
        SKILL_DOCTOR_FAIL_ON: ${{ inputs['fail-on'] }}
      run: |
        args=(
          "$SKILL_DOCTOR_PATH"
          "--format" "$SKILL_DOCTOR_FORMAT"
          "--fail-on" "$SKILL_DOCTOR_FAIL_ON"
        )
        if [[ -n "$SKILL_DOCTOR_OUTPUT" ]]; then
          args+=("--output" "$SKILL_DOCTOR_OUTPUT")
        fi
        skill-doctor "${args[@]}"