Agent skill

codebase-architect

Use when generating exhaustive architecture documentation for any WordPress plugin/theme with built-in self-checking, progress tracking, and verification loops. Produces comprehensive docs covering classes, functions, hooks, REST API, AJAX, JS, DB, templates, blocks, and more.

Stars 163
Forks 31

Install this agent skill to your Project

npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/codebase-architect-vapvarun-claude-backup

SKILL.md

Codebase Architect

Generate exhaustive architecture documentation for any WordPress plugin/theme with built-in self-checking, progress tracking, and verification loops.

When to use

Use this skill when:

  • Creating comprehensive architecture documentation for a plugin/theme
  • Auditing what exists in a codebase (classes, functions, hooks, APIs)
  • Verifying existing documentation completeness
  • Onboarding to a new codebase and need a reference guide

Inputs required

  • Path to the plugin or theme directory.
  • Scope level: architectural, hybrid, reference, or verify.
  • Output directory for documentation files.

Scope Parameter

Scope Coverage Best For Output Size
architectural Conceptual 95%, Reference 15% Large codebases (500+ files) 2,000-4,000 lines
hybrid Conceptual 95%, Reference 50% Medium codebases (50-500 files) 4,000-8,000 lines
reference Conceptual 95%, Reference 90%+ Small codebases (<50 files) 8,000+ lines
verify Verification only Check existing docs Report only

Procedure

0) Initialize manifest

Create manifest directory and tracking files:

bash
mkdir -p <output-dir>/manifest

Create manifest files for each category:

  • classes.txt, functions.txt, hooks.txt
  • rest-endpoints.txt, ajax-handlers.txt
  • js-files.txt, db-tables.txt, templates.txt
  • blocks.txt, shortcodes.txt, widgets.txt
  • cpt-taxonomies.txt, admin-pages.txt
  • cron-jobs.txt, cli-commands.txt
  • PROGRESS.md (master tracker)

Read:

  • references/manifest-format.md

1) Run enumeration (parallel agents)

Launch 16 enumeration agents in parallel to populate manifests:

  1. Classes: grep -rn "^class \|^abstract class \|^trait \|^interface "
  2. Functions: grep -rn "^function [a-z_]"
  3. Action hooks: grep -rohn "do_action\s*(\s*['\"][^'\"]*['\"]"
  4. Filter hooks: grep -rohn "apply_filters\s*(\s*['\"][^'\"]*['\"]"
  5. REST endpoints: grep -rn "register_rest_route"
  6. AJAX handlers: grep -rn "wp_ajax_"
  7. JavaScript files: find -name "*.js"
  8. DB tables: grep -rn "CREATE TABLE"
  9. Templates: find -name "*.php" -path "*/templates/*"
  10. Blocks: find -name "block.json"
  11. Shortcodes: grep -rn "add_shortcode"
  12. Widgets: grep -rn "extends WP_Widget"
  13. CPT/Taxonomies: grep -rn "register_post_type\|register_taxonomy"
  14. Admin pages: grep -rn "add_menu_page\|add_submenu_page"
  15. Cron jobs: grep -rn "wp_schedule_event"
  16. CLI commands: grep -rn "WP_CLI::add_command"

Read:

  • references/enumeration-commands.md

2) Document (with progress updates)

For each category, launch documentation agents that:

  1. Read from manifest
  2. Document each item
  3. Update manifest status: pending → documented
  4. Update PROGRESS.md

Read:

  • references/documentation-agents.md

3) Self-verification loop

After documentation completes:

  1. Count items with status:pending (gaps)
  2. Count items with status:documented (done)
  3. If coverage < 100%, trigger gap filling
  4. Run cross-reference check

Read:

  • references/verification-loop.md

4) Gap filling (automatic)

For any gaps found:

  1. Read source code for item
  2. Generate documentation
  3. Update manifest
  4. Trigger verification again

5) Final validation

When verification passes:

  • Structure check (TOC, links, code examples)
  • Completeness check (all categories at 100%)
  • Quality check (each item has required fields)
  • Generate FINAL_REPORT.md

Verification

Documentation is COMPLETE when:

  • All 15 manifest categories at 100% coverage
  • Verification loop passes without finding new gaps
  • Cross-reference check passes
  • Final validation checklist complete
  • FINAL_REPORT.md generated

Failure modes / debugging

  • Enumeration returns 0 items:
    • Path doesn't exist, wrong grep pattern, vendor folder included
  • Documentation incomplete after multiple passes:
    • Check for status:error items in manifests
    • Review Gap Log in PROGRESS.md
  • Cross-reference failures:
    • Stale references from renamed/moved files

Read:

  • references/debugging.md

Escalation

  • For very large codebases (1000+ files), use architectural scope
  • If enumeration takes too long, exclude vendor/node_modules
  • For manual review items, document what's known and mark as partial

Output Files

<output-dir>/
├── PLUGIN_ARCHITECTURE.md  # Final documentation
├── manifest/
│   ├── PROGRESS.md         # Progress tracker
│   ├── classes.txt
│   ├── functions.txt
│   └── ... (other manifests)
└── FINAL_REPORT.md         # Validation report

Expand your agent's capabilities with these related and highly-rated skills.

Didn't find tool you were looking for?

Be as detailed as possible for better results