validate-docs-implementation

Validate Documentation vs Implementation

Safety Notice

This listing is imported from skills.sh public index metadata. Review upstream SKILL.md and repository scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "validate-docs-implementation" with this command: npx skills add abdullahbeam/nexus-design-abdullah/abdullahbeam-nexus-design-abdullah-validate-docs-implementation

Validate Documentation vs Implementation

Purpose: Systematically identify and fix mismatches between actual implementation code and documentation.

Load When:

  • User says: "validate docs", "check documentation", "docs vs implementation"

  • User mentions: "documentation mismatch", "docs don't match code"

  • After major code changes that may affect docs

  • During pre-release validation

Core Value: Ensures documentation accurately reflects actual implementation, preventing user confusion and wasted debugging time.

Quick Reference

What This Skill Does:

  • ✅ Analyzes implementation code (scripts, core logic)

  • ✅ Identifies what the code actually creates/does

  • ✅ Searches all documentation for references to those entities

  • ✅ Compares documented behavior vs actual behavior

  • ✅ Fixes all mismatches systematically

  • ✅ Provides summary of changes made

Typical Mismatches Found:

  • File names (e.g., docs say "design.md" but code creates "plan.md")

  • Folder structure (e.g., docs missing "02-resources/" directory)

  • Counts (e.g., docs say "3 files" but script creates "4 directories + 3 files")

  • Naming conventions (e.g., old vs new terminology)

  • References to deprecated features

Workflow: 5-Step Validation Process

Step 1: Analyze Implementation

Goal: Understand what the code actually does

Actions:

  • Read the implementation file(s) identified by user

  • Document exactly what it creates/generates/does:

  • Files created (names, locations, count)

  • Folders created (names, structure, count)

  • Data structures used

  • Behavior patterns

  • Create a reference list of "ground truth" facts

Example Output:

Implementation Analysis: init_project.py

Creates:

  • 4 directories: 01-planning/, 02-resources/, 03-working/, 04-outputs/
  • 3 files in 01-planning/: overview.md, plan.md, steps.md

Does NOT create:

  • design.md (old name)
  • tasks.md (old name)
  • requirements.md (never existed)

Step 2: Search Documentation

Goal: Find all references to the entities/behavior

Actions:

Use Grep to search for old/mismatched terms across documentation:

Search for old file names

grep -r "design.md|tasks.md|requirements.md" 00-system/

Search for old folder structure

grep -r "02-working|03-outputs" 00-system/

Identify all files that contain references

For each file, note line numbers and context

Output: List of files needing updates with specific locations

Step 3: Compare & Identify Mismatches

Goal: Categorize what needs fixing

Actions:

For each documentation file found:

  • Read the relevant sections

  • Compare against implementation truth

  • Categorize mismatch type:

  • File name mismatch: Wrong file names referenced

  • Structure mismatch: Wrong folder structure shown

  • Count mismatch: Wrong counts ("3 files" vs "4 directories + 3 files")

  • Missing info: Implementation creates X but docs don't mention it

Prioritize fixes:

  • Critical: User-facing docs, core system maps

  • Important: Skill workflows, reference materials

  • Nice-to-have: Examples, comments

Step 4: Fix All Mismatches

Goal: Update documentation to match implementation

Actions:

For each file with mismatches:

  • Read the file

  • Use Edit tool to fix each mismatch

  • Preserve context and formatting

  • Update related references (e.g., if fixing folder name, update all mentions)

Common fix patterns:

File name fixes

OLD: "design.md" → NEW: "plan.md" OLD: "tasks.md" → NEW: "steps.md"

Structure fixes

OLD: "02-working/" → NEW: "03-working/" OLD: "03-outputs/" → NEW: "04-outputs/" ADD: "02-resources/" (was missing)

Count fixes

OLD: "3 core files" NEW: "4 directories (01-planning/, 02-resources/, 03-working/, 04-outputs/) + 3 planning files"

Track progress with TodoWrite:

  • Mark each file as completed after fixing

  • Maintains visibility for user

Step 5: Verify & Generate Report

Goal: Confirm all fixes are consistent and create permanent validation record

Actions:

5A. Verification:

Re-search for old terms to verify they're gone:

grep -r "design.md" 00-system/ # Should return no results

Spot-check fixed files to ensure consistency

5B. Determine Report Location:

First time or if location not set:

AI: "Where should I save validation reports? Options: 1. 04-workspace/validation-reports/ (recommended - organized, searchable) 2. 01-memory/validation-reports/ (persistent record) 3. Custom location (you specify)

Or say 'skip' to not save reports."

User: [Chooses option]

AI: "Got it! I'll save reports to [location]. This will be remembered for future validations."

If location already configured (from previous validation):

AI: "Saving report to configured location: [location]"

5C. Generate & Save Report:

  • Create validation report using template (see references/report-template.md )

  • Generate filename: validation-{implementation}-YYYY-MM-DD.md

  • Example: validation-init-project-2025-11-24.md

  • Save to configured location

  • Display summary to user with link to full report

Report includes:

  • Implementation analyzed

  • Files updated (with line-by-line changes)

  • Mismatch breakdown by type

  • Verification results

  • Status and next steps

Example output:

✅ Validation complete!

Summary:

  • 6 files updated
  • 23 total fixes
  • 0 grep results for old terms

📄 Full report saved: 04-workspace/validation-reports/validation-init-project-2025-11-24.md

Common Validation Scenarios

Scenario 1: Script Creates Different Files Than Documented

Symptoms:

  • User reports "docs mention design.md but I don't see it"

  • Script creates files not mentioned in docs

Process:

  • Read script to see what it actually creates

  • Search docs for old file names

  • Replace all references systematically

Example: The create-project validation we just did

Scenario 2: Folder Structure Changed

Symptoms:

  • Documentation shows old folder structure

  • New folders not mentioned in docs

Process:

  • Check actual project structure (ls)

  • Compare against documented structure

  • Update all structure diagrams and references

Scenario 3: Counts/Descriptions Wrong

Symptoms:

  • "Creates 3 files" but actually creates more

  • Missing important details about what's created

Process:

  • Count actual artifacts created

  • Find all count references in docs

  • Update with complete, accurate descriptions

Best Practices

Be Systematic

  • Don't skip steps or assume you've found everything

  • Use grep to search exhaustively

  • Check related files (if you fix system-map.md, check framework-overview.md too)

Preserve Context

  • Don't just fix the immediate error

  • Update related references nearby

  • Ensure examples/diagrams stay consistent

Use Parallel Searches

  • When searching for multiple patterns, use multiple Grep calls in parallel

  • Saves time and provides complete picture faster

Track Progress

  • Use TodoWrite to show user what's being fixed

  • Mark items complete as you go

  • Provides transparency and confidence

Verify Fixes

  • Re-search for old terms after fixing

  • Spot-check that new terminology is consistent

  • Run validation commands if available

Integration Points

  • After code changes: Run this validation before committing

  • Pre-release: Validate all docs match current implementation

  • User reports mismatch: Use this skill to fix comprehensively

  • Documentation updates: Verify changes didn't introduce new mismatches

Success Criteria

✅ Zero grep results for old/deprecated terms ✅ All structure diagrams match actual folder structure ✅ All counts match actual artifact counts ✅ All file references use current names ✅ User can follow docs without confusion

Remember: Documentation debt compounds quickly. Fix mismatches systematically and completely the first time!

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Related Skills

Related by shared tags or category signals.

General

mental-models

No summary provided by upstream source.

Repository SourceNeeds Review
General

notion-connect

No summary provided by upstream source.

Repository SourceNeeds Review
General

google-tasks

No summary provided by upstream source.

Repository SourceNeeds Review