feat(at-mention): implement :ClaudeCodeSend command for visual sele…
#1
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| name: Documentation Review | |
| on: | |
| push: | |
| branches: [main] | |
| paths: | |
| - "**.lua" | |
| - ".github/**" | |
| - "plugin/**" | |
| permissions: | |
| contents: write | |
| pull-requests: write | |
| jobs: | |
| review-docs: | |
| runs-on: ubuntu-latest | |
| steps: | |
| - name: Checkout repository | |
| uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4 | |
| with: | |
| fetch-depth: 0 # Fetch full history to analyze changes | |
| persist-credentials: true | |
| - name: Run Claude Code to review documentation | |
| id: claude-review | |
| uses: anthropics/claude-code-action@8b9d5bb25a638c9aa5103496c0139d99b4936d42 # beta | |
| with: | |
| anthropic_api_key: ${{ secrets.ANTHROPIC_API_KEY }} | |
| prompt: | | |
| As a Documentation Validator, your task is to review the current code in this repository and ensure that the markdown documentation files (README.md, ARCHITECTURE.md, and DEVELOPMENT.md) are up-to-date and accurate. | |
| Focus on these specific areas: | |
| 1. Configuration options - Ensure all config options in lua/claudecode/config.lua are accurately reflected in README.md | |
| 2. Commands and usage - Verify all commands and keymaps mentioned in plugin/claudecode.lua are correctly documented in README.md | |
| 3. Architecture details - Validate the module structure in ARCHITECTURE.md matches the actual codebase structure | |
| 4. Development status - Review the implementation status in DEVELOPMENT.md against the actual codebase | |
| 5. Terminal feature - Check that the terminal integration is documented correctly across all files | |
| For any inconsistencies found: | |
| 1. Note the specific issue (what's incorrect, missing, or outdated) | |
| 2. Provide the correct information based on the current codebase | |
| 3. Create a corrected version of the relevant section | |
| Output your findings in this format: | |
| ```json | |
| { | |
| "review_date": "YYYY-MM-DD", | |
| "files_checked": ["README.md", "ARCHITECTURE.md", "DEVELOPMENT.md"], | |
| "issues_found": [ | |
| { | |
| "file": "path/to/file.md", | |
| "section": "Section title", | |
| "issue": "Description of inconsistency", | |
| "corrected_content": "The corrected markdown content" | |
| } | |
| ] | |
| } | |
| ``` | |
| If no issues are found, return an empty "issues_found" array. | |
| - name: Process documentation updates | |
| id: process-updates | |
| shell: bash | |
| # zizmor-disable-next-line template-injection | |
| env: | |
| CLAUDE_OUTPUT: ${{ steps.claude-review.outputs.result }} | |
| run: | | |
| # Extract issues from Claude's output | |
| # Check if the output contains the expected JSON format | |
| if ! echo "$CLAUDE_OUTPUT" | grep -q "\"issues_found\""; then | |
| echo "No valid JSON output found from Claude review" | |
| exit 0 | |
| fi | |
| # Extract the JSON part (between ```json and ```) | |
| echo "$CLAUDE_OUTPUT" | sed -n "/\`\`\`json/,/\`\`\`/p" | sed "1d;\$d" > json_output.txt | |
| echo "JSON output extracted" | |
| # Parse the JSON and check if there are issues | |
| ISSUES_FOUND=$(jq ".issues_found | length" json_output.txt) | |
| echo "Issues found: $ISSUES_FOUND" | |
| # Clean up json_output.txt if no issues are found or after processing | |
| if [ "$ISSUES_FOUND" -eq 0 ]; then | |
| echo "No documentation issues found" | |
| rm -f json_output.txt # Clean up if exiting early | |
| exit 0 | |
| fi | |
| if [ "$ISSUES_FOUND" -eq 0 ]; then | |
| echo "No documentation issues found" | |
| exit 0 | |
| fi | |
| echo "Found $ISSUES_FOUND documentation issues to fix" | |
| echo "issues_found=$ISSUES_FOUND" >> "$GITHUB_OUTPUT" | |
| # Create a new branch for the fixes | |
| BRANCH_NAME="docs/update-$(date +%Y%m%d%H%M%S)" | |
| git checkout -b "$BRANCH_NAME" | |
| # Apply each correction - use a safer approach | |
| for i in $(seq 0 $((ISSUES_FOUND - 1))); do | |
| # Extract each field safely into separate files | |
| jq -r ".issues_found[$i].file" json_output.txt > issue_file.txt | |
| jq -r ".issues_found[$i].section" json_output.txt > issue_section.txt | |
| jq -r ".issues_found[$i].issue" json_output.txt > issue_desc.txt | |
| jq -r ".issues_found[$i].corrected_content" json_output.txt > issue_content.txt | |
| # Read from files | |
| FILE=$(cat issue_file.txt) | |
| SECTION=$(cat issue_section.txt) | |
| ISSUE=$(cat issue_desc.txt) | |
| CONTENT=$(cat issue_content.txt) | |
| echo "Updating $FILE - Section: $SECTION" | |
| # Escape content for sed using files for safety | |
| sed -E "s/[\/&]/\\&/g" issue_content.txt > escaped_content.txt | |
| ESCAPED_CONTENT=$(cat escaped_content.txt) | |
| # Look for the section and apply the change | |
| # This is a basic approach; more sophisticated section matching might be needed | |
| if grep -q "$SECTION" "$FILE"; then | |
| # Try to find the section and replace content from the section title until the next section | |
| sed -i "/# $SECTION/,/# /c\\# $SECTION\n\n$ESCAPED_CONTENT\n\n# " "$FILE" | |
| else | |
| echo "Section not found: $SECTION in $FILE. Adding to the end of the file." | |
| echo -e "\n# $SECTION\n\n$CONTENT" >> "$FILE" | |
| fi | |
| done | |
| # Clean up temporary files used in the loop | |
| rm -f issue_file.txt issue_section.txt issue_desc.txt issue_content.txt escaped_content.txt | |
| # Clean up json_output.txt after it's fully processed | |
| rm -f json_output.txt | |
| # Set up git identity for Claude | |
| git config --local user.email "[email protected]" | |
| git config --local user.name "Claude" | |
| # Commit the changes | |
| git add README.md ARCHITECTURE.md DEVELOPMENT.md | |
| git commit -m "docs: update documentation based on code analysis" | |
| git push origin "$BRANCH_NAME" | |
| echo "branch_name=$BRANCH_NAME" >> "$GITHUB_OUTPUT" | |
| # Create summary of changes to display in PR | |
| CHANGES_SUMMARY="" | |
| for i in $(seq 0 $((ISSUES_FOUND - 1))); do | |
| FILE=$(echo "$JSON_OUTPUT" | jq -r ".issues_found[$i].file") | |
| SECTION=$(echo "$JSON_OUTPUT" | jq -r ".issues_found[$i].section") | |
| ISSUE=$(echo "$JSON_OUTPUT" | jq -r ".issues_found[$i].issue") | |
| CHANGES_SUMMARY="${CHANGES_SUMMARY}- **${FILE}** (${SECTION}): ${ISSUE}\n" | |
| done | |
| { | |
| echo "changes_summary<<EOF" | |
| echo -e "$CHANGES_SUMMARY" | |
| echo "EOF" | |
| } >> "$GITHUB_OUTPUT" | |
| - name: Create Pull Request | |
| if: steps.process-updates.outputs.issues_found > 0 | |
| uses: anthropics/claude-pr-action@8b9d5bb25a638c9aa5103496c0139d99b4936d42 # v1 | |
| with: | |
| anthropic_api_key: ${{ secrets.ANTHROPIC_API_KEY }} | |
| github_token: ${{ secrets.GITHUB_TOKEN }} | |
| branch: ${{ steps.process-updates.outputs.branch_name }} | |
| base: main | |
| title: "docs: Update documentation to match current codebase" | |
| body: | | |
| This PR was automatically generated by Claude Code to update documentation files based on code analysis. | |
| ## What's Changed | |
| Documentation has been updated to match the current codebase. The following issues were addressed: | |
| ${{ steps.process-updates.outputs.changes_summary }} | |
| The following files were updated: | |
| - README.md | |
| - ARCHITECTURE.md | |
| - DEVELOPMENT.md | |
| Please review these changes to ensure they accurately reflect the current state of the project. |