Claude Research

Research documentation for: $ARGUMENTS

System Prompt

⚠️ EXECUTION DIRECTIVE: When the user invokes this command, you MUST:

IMMEDIATELY execute - no questions, no explanations first
ALWAYS show full output from script/tool execution
ALWAYS provide summary after execution completes

DO NOT:

Say "I don't need to take action" - you ALWAYS execute when invoked
Ask for confirmation unless requires_confirmation: true in frontmatter
Skip execution even if it seems redundant - run it anyway

The user invoking the command IS the confirmation.

Arguments

$ARGUMENTS - Research Scope (optional)

Specify what to research:

Batch processing: No arguments or "--batch" - processes all NEW/ files
Specific URL: "Research https://docs.anthropic.com/claude-code/installation"
Category focus: "Research official API documentation"
Watch mode: "Monitor NEW/ directory continuously"

Default Behavior

If no arguments:

Processes files in docs/original-research/NEW/
Converts and categorizes content
Archives sources to BU/
Updates category indexes
Commits changes

Overview

Automated research and organization of Claude Code and Anthropic documentation from official sources, community resources, and training materials.

Syntax

/claude-research [options]

Options

Source Selection

--sources <source-type>
  official          # Anthropic official documentation only
  community         # Community content (GitHub, Reddit, blogs)
  training          # Training materials and courses
  all               # All sources (default)

--url <url>         # Process specific URL
--file <path>       # Process specific local file
--batch             # Process all files in NEW/ directory

Category Targeting

--categories <categories>
  api               # API reference documentation
  tutorials         # Step-by-step tutorials
  best-practices    # Best practice guides
  releases          # Release notes and updates
  blogs             # Blog posts and articles
  discussions       # Community discussions

# Multiple categories (comma-separated)
--categories "api,tutorials,best-practices"

Processing Options

--format <format>
  markdown          # Output as markdown (default)
  pdf               # Save as PDF
  html              # Save as HTML

--duplicate-check   # Enable duplicate detection (default: enabled)
--no-duplicate-check # Disable duplicate detection

--quality-check     # Enable quality validation (default: enabled)
--no-quality-check  # Skip quality validation

Automation Options

--auto-commit       # Automatically git commit processed files
--auto-push         # Automatically git push after commit
--watch             # Continuous monitoring of NEW/ directory
--notify            # Send notification on completion

Output Control

--output-dir <path> # Custom output directory (default: docs/research-library/)
--archive-dir <path> # Custom archive directory (default: docs/original-research/BU/)
--dry-run           # Show what would be done without executing
--verbose           # Detailed output logging

Usage Examples

Example 1: Process New Files in NEW/ Directory

/claude-research --batch --auto-commit

What it does:

Finds all files in docs/original-research/NEW/
Processes each file (convert, categorize, organize)
Archives sources to BU/
Updates category indexes
Commits with message: docs: Add [count] new research documents

Example 2: Scrape Official Anthropic Documentation

/claude-research --sources official --categories "api,tutorials" --auto-commit

What it does:

Scrapes docs.anthropic.com for API and tutorial documentation
Converts to markdown format
Organizes into docs/research-library/official/api/ and .../tutorials/
Creates/updates category indexes
Commits changes

Example 3: Process Specific URL

/claude-research --url "https://docs.anthropic.com/claude-code/installation" \
                 --categories tutorials --auto-commit

What it does:

Fetches content from URL
Converts to markdown
Saves to docs/research-library/official/tutorials/installation.md
Archives source HTML
Updates indexes and commits

Example 4: Watch Mode for Continuous Processing

/claude-research --watch --auto-commit --notify

What it does:

Monitors docs/original-research/NEW/ for new files
Auto-processes when files are added
Commits each processed file
Sends notification on completion

Example 5: Dry Run (Preview Only)

/claude-research --batch --dry-run --verbose

What it does:

Scans NEW/ directory
Shows what would be processed
Displays categorization decisions
Shows target paths
Does NOT execute moves or commits

Example 6: Custom Output Directory

/claude-research --file "my-transcript.txt" \
                 --output-dir "docs/custom-research/" \
                 --categories tutorials

What it does:

Processes my-transcript.txt
Saves to docs/custom-research/tutorials/
Archives source as usual
Updates custom directory index

Workflow Details

Standard Processing Workflow

1. Discovery
   └─ Scan NEW/ or fetch URL

2. Reading
   └─ Load file content or web page

3. Processing
   ├─ Convert transcript to markdown (if .txt)
   ├─ Extract metadata (title, source, date)
   └─ Format with proper structure

4. Categorization
   ├─ Analyze content
   ├─ Determine category (official/community/training/releases)
   └─ Select subdirectory (api/tutorials/blogs/etc.)

5. Duplicate Detection
   ├─ Search existing content
   ├─ Compare similarity
   └─ Flag duplicates or skip

6. Organization
   ├─ Move to target directory: docs/research-library/{category}/
   ├─ Archive source to: docs/original-research/BU/
   └─ Update category README.md index

7. Version Control
   ├─ git add processed files
   ├─ git commit -m "docs: Add [filename] to [category]"
   └─ git push (if --auto-push)

8. Reporting
   └─ Generate processing summary

Watch Mode Workflow

while monitoring:
    on file_added to NEW/:
        process_file(file)
        archive_source(file)
        git_commit()
        notify_completion()

Output Structure

Processed files are organized into:

docs/research-library/
├── README.md                           # Master index
├── official/
│   ├── README.md
│   ├── api/
│   │   ├── README.md
│   │   └── [processed-docs].md
│   ├── tutorials/
│   │   ├── README.md
│   │   └── [processed-docs].md
│   └── best-practices/
│       ├── README.md
│       └── [processed-docs].md
├── community/
│   ├── README.md
│   ├── blogs/
│   │   ├── README.md
│   │   └── [processed-docs].md
│   └── discussions/
│       ├── README.md
│       └── [processed-docs].md
├── training/
│   ├── README.md
│   └── courses/
│       ├── README.md
│       └── [processed-docs].md
└── releases/
    ├── README.md
    └── version-2.0/
        ├── README.md
        └── [processed-docs].md

Category Index Format

Each category directory contains a README.md index:

# [Category Name]

**Last Updated:** 2025-11-29
**Total Documents:** 15

## Documents

### API Reference
- [Claude Code API](api-reference.md) - Complete API documentation
- [Authentication](authentication.md) - Auth patterns and examples

### Tutorials
- [Getting Started](getting-started.md) - Beginner guide
- [Advanced Workflows](advanced-workflows.md) - Expert techniques

## Recent Additions
- 2025-11-29: [Authentication](authentication.md)
- 2025-11-28: [Getting Started](getting-started.md)

Master Index Format

docs/research-library/README.md provides complete inventory:

# Claude Code Research Plan

**Last Updated:** 2025-11-29
**Total Documents:** 47
**Categories:** 7

## Quick Navigation

- [Official Documentation](#official-documentation) (25 docs)
- [Community Content](#community-content) (12 docs)
- [Training Materials](#training-materials) (5 docs)
- [Release Notes](#release-notes) (5 docs)

## Official Documentation

### API Reference (8 docs)
- See [official/api/README.md](official/api/README.md)

### Tutorials (12 docs)
- See [official/tutorials/README.md](official/tutorials/README.md)

### Best Practices (5 docs)
- See [official/best-practices/README.md](official/best-practices/README.md)

## Community Content

### Blogs (7 docs)
- See [community/blogs/README.md](community/blogs/README.md)

### Discussions (5 docs)
- See [community/discussions/README.md](community/discussions/README.md)

## Training Materials

### Courses (5 docs)
- See [training/courses/README.md](training/courses/README.md)

## Release Notes

### Version 2.0 (5 docs)
- See [releases/version-2.0/README.md](releases/version-2.0/README.md)

## Statistics

- Most active category: Official/Tutorials (12 docs)
- Most recent addition: 2025-11-29
- Growth rate: +3 docs/week (average)

Commit Message Format

Uses conventional commit format:

docs: Add [filename] to [category]

- Source: [URL or file path]
- Category: [category/subdirectory]
- Type: [API/Tutorial/Blog/etc.]
- Date: [YYYY-MM-DD]

Processed by claude-research-agent

Examples:

docs: Add Claude Code installation guide to official/tutorials

- Source: https://docs.anthropic.com/claude-code/installation
- Category: official/tutorials
- Type: Tutorial
- Date: 2025-11-29

Processed by claude-research-agent

docs: Add 3 new community blog posts

- community/blogs/advanced-prompting.md
- community/blogs/multi-instance-workflows.md
- community/blogs/mcp-integration-guide.md

Batch processed by claude-research-agent

Error Handling

Duplicate Content:

⚠️  Duplicate detected: installation-guide.md
    Existing: docs/research-library/official/tutorials/installation-guide.md
    Action: Skipped processing
    Recommendation: Review content differences manually

Invalid Format:

❌ Format error: corrupt-file.txt
    Issue: Unable to parse markdown structure
    Action: Saved to docs/research-library/manual-review/
    Recommendation: Manual processing required

Category Ambiguity:

⚠️  Category unclear: hybrid-content.md
    Confidence: official/tutorials (65%), community/blogs (35%)
    Action: Defaulted to official/tutorials
    Recommendation: Review categorization

Git Commit Failure:

❌ Git commit failed
    Error: Unable to push to remote
    Action: Changes committed locally
    Command: cd /path/to/repo && git push

Performance Metrics

Command provides performance report:

📊 Processing Summary

Files Processed: 5
Time Elapsed: 47 seconds
Average Speed: 9.4 seconds/file

Categorization:
├─ official/tutorials: 3 files
├─ community/blogs: 1 file
└─ training/courses: 1 file

Quality Checks:
├─ Passed: 5 files (100%)
├─ Warnings: 1 file (20%)
└─ Failed: 0 files (0%)

Duplicates:
├─ Detected: 2 files
└─ Skipped: 2 files

Git Operations:
├─ Commits: 1
├─ Files staged: 7
└─ Push: Success

Configuration

Settings stored in .coditect/settings.json:

{
  "commands": {
    "claude-research": {
      "default_sources": "all",
      "default_categories": ["api", "tutorials", "best-practices"],
      "output_directory": "docs/research-library/",
      "archive_directory": "docs/original-research/BU/",
      "auto_commit": true,
      "auto_push": false,
      "duplicate_check": true,
      "quality_check": true,
      "watch_mode_interval": 5,
      "notification": {
        "enabled": false,
        "method": "slack",
        "webhook_url": ""
      }
    }
  }
}

Integration with Other Commands

Complementary Commands:

# Process research, then export for deduplication
/claude-research --batch --auto-commit
/export-dedup

# Research official docs, then analyze patterns
/claude-research --sources official --categories api
/analyze-session

# Batch process and create git workflow
/claude-research --batch
/git-sync --target all

Dependencies

Required Components:

claude-research-agent (this command's backend)
codi-documentation-writer (for markdown formatting)
web-search-researcher (for web scraping)

Optional Enhancements:

git-workflow-orchestrator (for advanced git automation)
project-organizer (for directory maintenance)

Activation Instructions

Status: NOT ACTIVATED

To activate this command:

cd /Users/halcasteel/Downloads/CLAUDE-CODE-HOWTOs/.coditect
python3 scripts/update-component-activation.py activate command claude-research \
  --reason "Automated Claude/Anthropic documentation research and organization"

git add commands/claude-research.md .coditect/component-activation-status.json
git commit -m "feat(command): Add /claude-research slash command"
git push

Verify activation:

python3 scripts/update-component-activation.py status command claude-research

Troubleshooting

Command not recognized:

Verify component activation status
Check .coditect/component-activation-status.json
Restart Claude Code session

Files not processing:

Check NEW/ directory permissions
Verify file format (.txt or .md)
Review error logs in verbose mode

Categorization incorrect:

Review categorization logic in agent
Manually move files to correct category
Update category index manually

Git commit fails:

Check git repository status
Verify remote connection
Commit manually and report issue

Success Output

When research processing completes:

✅ COMMAND COMPLETE: /claude-research
Files Processed: N
Categories: <list>
Archive: docs/original-research/BU/
Output: docs/research-library/
Commits: N

Completion Checklist

Before marking complete:

Failure Indicators

This command has FAILED if:

❌ Source files not found
❌ Category unknown
❌ Processing failed
❌ No output generated

When NOT to Use

Do NOT use when:

Manual organization needed
Non-Claude documentation
Already categorized content

Anti-Patterns (Avoid)

Anti-Pattern	Problem	Solution
Skip categorization	Disorganized content	Always categorize
No duplicate check	Redundant content	Enable --duplicate-check
Ignore errors	Lost content	Review error logs

Principles

This command embodies:

#2 Search Before Create - Find existing content
#3 Complete Execution - Full processing workflow
#6 Clear, Understandable - Organized output

Full Standard: CODITECT-STANDARD-AUTOMATION.md

Command Version: 1.0.0 Agent Framework: Universal Agent Framework v2.0 Compliance: CODITECT Component Standards v1.0 Last Updated: 2025-11-29

System Prompt​

Arguments​

$ARGUMENTS - Research Scope (optional)​

Default Behavior​

Overview​

Syntax​

Options​

Source Selection​

Category Targeting​

Processing Options​

Automation Options​

Output Control​

Usage Examples​

Example 1: Process New Files in NEW/ Directory​

Example 2: Scrape Official Anthropic Documentation​

Example 3: Process Specific URL​

Example 4: Watch Mode for Continuous Processing​

Example 5: Dry Run (Preview Only)​

Example 6: Custom Output Directory​

Workflow Details​

Standard Processing Workflow​

Watch Mode Workflow​

Output Structure​

Category Index Format​

Master Index Format​

Commit Message Format​

Error Handling​

Performance Metrics​

Configuration​

Integration with Other Commands​

Dependencies​

Activation Instructions​

Troubleshooting​

See Also​

Success Output​

Completion Checklist​

Failure Indicators​

When NOT to Use​

Anti-Patterns (Avoid)​

Principles​

System Prompt

Arguments

$ARGUMENTS - Research Scope (optional)

Default Behavior

Overview

Syntax

Options

Source Selection

Category Targeting

Processing Options

Automation Options

Output Control

Usage Examples

Example 1: Process New Files in NEW/ Directory

Example 2: Scrape Official Anthropic Documentation

Example 3: Process Specific URL

Example 4: Watch Mode for Continuous Processing

Example 5: Dry Run (Preview Only)

Example 6: Custom Output Directory

Workflow Details

Standard Processing Workflow

Watch Mode Workflow

Output Structure

Category Index Format

Master Index Format

Commit Message Format

Error Handling

Performance Metrics

Configuration

Integration with Other Commands

Dependencies

Activation Instructions

Troubleshooting

See Also

Success Output

Completion Checklist

Failure Indicators

When NOT to Use

Anti-Patterns (Avoid)

Principles