action.yml

name: 'Context Lint'
description: 'Lint a URL for LLM readiness and token efficiency'
author: 'Hansel Wahjono'

branding:
  icon: 'search'
  color: 'blue'

inputs:
  url:
    description: 'URL to lint'
    required: true
  fail-under:
    description: 'Fail if score is below this threshold (0-100)'
    required: false
  fail-on-blocked-bots:
    description: 'Fail (exit 2) if any AI bot is blocked'
    required: false
    default: 'false'
  single-page:
    description: 'Lint only the given URL (skip multi-page discovery)'
    required: false
    default: 'false'
  max-pages:
    description: 'Maximum pages to lint in multi-page mode'
    required: false
    default: '10'
  baseline-file:
    description: 'Path to baseline JSON file to compare against (regression detection)'
    required: false
  save-baseline:
    description: 'Path to save current scores as a baseline JSON file'
    required: false
  regression-threshold:
    description: 'Minimum score drop to flag as regression (default: 5 points)'
    required: false
    default: '5'
  robots-min:
    description: 'Minimum robots.txt pillar score (0-25)'
    required: false
  schema-min:
    description: 'Minimum Schema.org pillar score (0-25)'
    required: false
  content-min:
    description: 'Minimum content density pillar score (0-40)'
    required: false
  llms-min:
    description: 'Minimum llms.txt pillar score (0-10)'
    required: false
  overall-min:
    description: 'Minimum overall score (0-100, alternative to fail-under)'
    required: false
  webhook-url:
    description: 'Webhook URL to POST lint results to (Slack, Discord, etc.)'
    required: false
  max-context-waste:
    description: 'Maximum acceptable context waste percentage (0-100). Fail if exceeded.'
    required: false
    default: ''
  require-llms-txt:
    description: 'Fail if llms.txt is not present on the site'
    required: false
    default: 'false'
  require-bot-access:
    description: 'Fail if any AI bots are blocked by robots.txt'
    required: false
    default: 'false'
  python-version:
    description: 'Python version to use'
    required: false
    default: '3.12'

outputs:
  score:
    description: 'Overall LLM readiness score (0-100)'
    value: ${{ steps.lint.outputs.score }}
  report-json:
    description: 'Full lint report as JSON'
    value: ${{ steps.lint.outputs.report_json }}
  token-waste:
    description: 'Context waste percentage from Token Waste analysis'
    value: ${{ steps.lint.outputs.token_waste }}

runs:
  using: 'composite'
  steps:
    - name: Set up Python
      uses: actions/setup-python@v5
      with:
        python-version: ${{ inputs.python-version }}
        cache: 'pip'

    - name: Install context-cli
      shell: bash
      run: pip install context-linter

    - name: Install crawl4ai browser
      shell: bash
      run: |
        crawl4ai-setup || echo "::warning::crawl4ai browser setup failed, content analysis may be limited"

    - name: Run Context Lint
      id: lint
      shell: bash
      run: |
        # Build command
        CMD="context-cli lint '${{ inputs.url }}'"

        if [ "${{ inputs.single-page }}" = "true" ]; then
          CMD="$CMD --single"
        else
          CMD="$CMD --max-pages ${{ inputs.max-pages }}"
        fi

        CMD="$CMD --json"

        # Run lint and capture JSON output
        REPORT=$(eval $CMD 2>/dev/null || true)

        # Extract score from JSON
        SCORE=$(echo "$REPORT" | python3 -c "import sys, json; print(json.load(sys.stdin)['overall_score'])" 2>/dev/null || echo "0")

        # Extract token waste from JSON
        TOKEN_WASTE=$(echo "$REPORT" | python3 -c "
import sys, json
data = json.load(sys.stdin)
lr = data.get('lint_result')
if lr:
    print(lr.get('context_waste_pct', 'N/A'))
else:
    print('N/A')
" 2>/dev/null || echo "N/A")

        echo "score=$SCORE" >> $GITHUB_OUTPUT
        echo "token_waste=$TOKEN_WASTE" >> $GITHUB_OUTPUT
        echo "report_json<<REPORT_EOF" >> $GITHUB_OUTPUT
        echo "$REPORT" >> $GITHUB_OUTPUT
        echo "REPORT_EOF" >> $GITHUB_OUTPUT

        # Now run with formatting for the step summary
        SUMMARY_CMD="context-cli lint '${{ inputs.url }}'"

        if [ "${{ inputs.single-page }}" = "true" ]; then
          SUMMARY_CMD="$SUMMARY_CMD --single"
        else
          SUMMARY_CMD="$SUMMARY_CMD --max-pages ${{ inputs.max-pages }}"
        fi

        if [ -n "${{ inputs.fail-under }}" ]; then
          SUMMARY_CMD="$SUMMARY_CMD --fail-under ${{ inputs.fail-under }}"
        fi

        if [ "${{ inputs.fail-on-blocked-bots }}" = "true" ]; then
          SUMMARY_CMD="$SUMMARY_CMD --fail-on-blocked-bots"
        fi

        if [ -n "${{ inputs.baseline-file }}" ]; then
          SUMMARY_CMD="$SUMMARY_CMD --baseline '${{ inputs.baseline-file }}'"
          SUMMARY_CMD="$SUMMARY_CMD --regression-threshold ${{ inputs.regression-threshold }}"
        fi

        if [ -n "${{ inputs.save-baseline }}" ]; then
          SUMMARY_CMD="$SUMMARY_CMD --save-baseline '${{ inputs.save-baseline }}'"
        fi

        if [ -n "${{ inputs.robots-min }}" ]; then
          SUMMARY_CMD="$SUMMARY_CMD --robots-min ${{ inputs.robots-min }}"
        fi

        if [ -n "${{ inputs.schema-min }}" ]; then
          SUMMARY_CMD="$SUMMARY_CMD --schema-min ${{ inputs.schema-min }}"
        fi

        if [ -n "${{ inputs.content-min }}" ]; then
          SUMMARY_CMD="$SUMMARY_CMD --content-min ${{ inputs.content-min }}"
        fi

        if [ -n "${{ inputs.llms-min }}" ]; then
          SUMMARY_CMD="$SUMMARY_CMD --llms-min ${{ inputs.llms-min }}"
        fi

        if [ -n "${{ inputs.overall-min }}" ]; then
          SUMMARY_CMD="$SUMMARY_CMD --overall-min ${{ inputs.overall-min }}"
        fi

        if [ -n "${{ inputs.webhook-url }}" ]; then
          SUMMARY_CMD="$SUMMARY_CMD --webhook '${{ inputs.webhook-url }}'"
        fi

        if [ -n "${{ inputs.max-context-waste }}" ]; then
          SUMMARY_CMD="$SUMMARY_CMD --max-context-waste ${{ inputs.max-context-waste }}"
        fi

        if [ "${{ inputs.require-llms-txt }}" = "true" ]; then
          SUMMARY_CMD="$SUMMARY_CMD --require-llms-txt"
        fi

        if [ "${{ inputs.require-bot-access }}" = "true" ]; then
          SUMMARY_CMD="$SUMMARY_CMD --require-bot-access"
        fi

        eval $SUMMARY_CMD