Skip to main content

/what - Component Discovery

Discover WHAT components exist, WHAT they can do, and WHERE they are located.

System Prompt

EXECUTION DIRECTIVE: When /what is invoked, you MUST:

  1. IMMEDIATELY query the component database - no questions first
  2. Search dynamically - NEVER use static/hardcoded lists
  3. Return current counts from database, not from memory
  4. Include location (WHERE) in component details

Execution Steps

cd /Users/halcasteel/PROJECTS/coditect-rollout-master/submodules/core/coditect-core
source .venv/bin/activate

# For /what (overview)
python3 scripts/component-indexer.py --stats

# For /what <type>
python3 scripts/component-indexer.py --type agent
python3 scripts/component-indexer.py --type command
python3 scripts/component-indexer.py --type skill

# For /what --search <query>
python3 scripts/component-indexer.py --search "query"

# For /what can <component>
python3 scripts/component-indexer.py --capability "component-name"

# For counts (source of truth)
cat config/component-counts.json

Usage

# Overview (WHAT exists)
/what # Component inventory summary
/what stats # Detailed statistics

# List by type (WHAT types exist)
/what agents # All 776 agents
/what commands # All 377 commands
/what skills # All 445 skills
/what scripts # All 581 scripts
/what hooks # All 118 hooks
/what workflows # All 1,153 workflows

# Search (WHAT matches)
/what --search "kubernetes" # Search all components
/what agents --search "security" # Search within type

# Component details (WHAT can it do, WHERE is it)
/what can <component> # Capabilities
/what does <component> # Function explanation
/what is <component> # Full details + location

Response Format

For /what (Overview)

┌─────────────────────────────────────────────────────────────┐
│ CODITECT Component Inventory │
│ (Source: config/component-counts.json) │
├─────────────────────────────────────────────────────────────┤
│ Type │ Count │ Description │
│ ───────────┼───────┼───────────────────────────────────── │
│ Agents │ 776 │ AI agents with specialized expertise │
│ Commands │ 377 │ Slash commands for tasks │
│ Skills │ 445 │ Reusable skill definitions │
│ Scripts │ 581 │ Automation scripts (py, sh) │
│ Hooks │ 118 │ Event triggers │
│ Workflows │ 1153 │ Executable workflows (yaml, json) │
│ ───────────┼───────┼───────────────────────────────────── │
│ TOTAL │ 3458 │ │
├─────────────────────────────────────────────────────────────┤
│ Commands: │
│ • /what <type> - List components by type │
│ • /what can <name> - Show capabilities │
│ • /what is <name> - Full details + location │
│ • /what --search X - Search all components │
└─────────────────────────────────────────────────────────────┘

For /what agents

┌─────────────────────────────────────────────────────────────┐
│ CODITECT Agents (776) │
│ Location: agents/*.md │
├─────────────────────────────────────────────────────────────┤
│ DEVELOPMENT │
│ • orchestrator - Multi-agent coordination │
│ • senior-architect - System design, architecture │
│ • code-reviewer - Code quality assessment │
│ • rust-expert-developer - Rust systems programming │
│ • frontend-react-typescript-expert - React/TypeScript │
│ ... │
│ │
│ DEVOPS │
│ • devops-engineer - CI/CD pipelines, automation │
│ • cloud-architect - Cloud infrastructure │
│ • k8s-statefulset-specialist - Kubernetes │
│ ... │
│ │
│ SECURITY │
│ • security-specialist - Security audits │
│ • penetration-testing-agent - Penetration testing │
│ ... │
├─────────────────────────────────────────────────────────────┤
│ Use /what can <agent> for capabilities │
│ Use /which <task> to find best agent for your task │
└─────────────────────────────────────────────────────────────┘

For /what can <component>

┌─────────────────────────────────────────────────────────────┐
│ /what can: orchestrator │
├─────────────────────────────────────────────────────────────┤
│ CAPABILITIES │
│ • Coordinate multiple agents for complex tasks │
│ • Decompose tasks into subtasks │
│ • Manage dependencies between tasks │
│ • Track progress across agents │
│ • Handle failures and retries │
│ │
│ TOOLS │
│ Read, Write, Edit, Bash, Grep, Glob, TodoWrite │
│ │
│ MODEL │
│ sonnet (default), opus (complex coordination) │
│ │
│ CAN ORCHESTRATE │
│ Yes - coordinates 123 other agents │
├─────────────────────────────────────────────────────────────┤
│ WHERE │
│ File: agents/orchestrator.md │
│ Full: /Users/.../coditect-core/agents/orchestrator.md │
├─────────────────────────────────────────────────────────────┤
│ Related: /how use orchestrator, /which multi-agent task │
└─────────────────────────────────────────────────────────────┘

For /what is <component>

┌─────────────────────────────────────────────────────────────┐
│ /what is: /classify │
├─────────────────────────────────────────────────────────────┤
│ TYPE: command │
│ VERSION: 1.2.0 │
│ STATUS: active │
│ │
│ WHAT IT DOES │
│ Classifies markdown documents using MoE (Mixture of │
│ Experts) with sentence-transformers for semantic analysis. │
│ Updates frontmatter with type, confidence, and timestamp. │
│ │
│ CAPABILITIES │
│ • Semantic document classification │
│ • 5-judge consensus system │
│ • Frontmatter auto-update │
│ • Batch processing │
│ │
│ WHERE │
│ Command: commands/classify.md │
│ Script: scripts/moe_classifier/classify.py │
│ Config: scripts/moe_classifier/config/ │
│ │
│ USAGE │
│ /classify path/to/file.md --update-frontmatter │
│ /classify path/to/dir/ --recursive │
├─────────────────────────────────────────────────────────────┤
│ Related: /how classify documents, /how does MoE │
└─────────────────────────────────────────────────────────────┘

Search Examples

/what --search "kubernetes"
# Returns: k8s-statefulset-specialist, container-orchestration skill,
# deploy-k8s.sh, kubernetes workflows...

/what agents --search "security"
# Returns: security-specialist, penetration-testing-agent,
# backend-api-security, multi-tenant-architect...

/what commands --search "git"
# Returns: /git-sync, /smart-merge, /commit, /pr-enhance...

Comparison: /what vs /which vs /how

CommandQuestionUse When
/whatWHAT exists? WHERE is it?Exploring available components
/whichWHO/WHICH for this task?Have a task, need best tool
/howHOW to accomplish?Need guidance on doing something
# Discovery flow
/what agents # "What agents exist?"
/what can orchestrator # "What can orchestrator do?"
/which coordinate tasks # "Which agent for coordination?"
/how use orchestrator # "How do I use orchestrator?"

CommandPurpose
/which <task>Find best agent for task
/how <task>Get guidance on task
/where <component>Alias for /what is <component>
/why <component>Alias for /how does <component>

Success Output

When discovery completes successfully:

✅ COMMAND COMPLETE: /what
Query: <type or search term>
Results: N components found
Source: config/component-counts.json

Completion Checklist

Before marking complete:

  • Database queried successfully
  • Results formatted and displayed
  • Counts shown are current
  • Related commands suggested

Failure Indicators

This command has FAILED if:

  • ❌ Component database not indexed
  • ❌ component-counts.json missing
  • ❌ Query returned error
  • ❌ No results for valid type

When NOT to Use

Do NOT use when:

  • Need to find best tool for task (use /which)
  • Need guidance on how to do something (use /how)
  • Already know which component to use

Anti-Patterns (Avoid)

Anti-PatternProblemSolution
Use static listsOutdated infoAlways query database
Ignore searchMiss componentsUse --search for specifics
Confuse with /whichWrong results/what=inventory, /which=recommendation

Principles

This command embodies:

  • #6 Clear, Understandable, Explainable - Output immediately understandable
  • #9 Based on Facts, Cross-Check - Queries database, not hardcoded data

Full Standard: CODITECT-STANDARD-AUTOMATION.md


Version: 2.0.0 Created: 2026-01-03 Updated: 2026-01-03 Author: CODITECT Team