Discovery

Discovery Skill

Code discovery combining file-based search with knowledge graph-powered semantic search.

Quick Reference

• Templates: See templates.md for reporting discovery results

Available Actions

search_files

Find files by name pattern.

Parameters:

• pattern (required): Glob pattern to match (e.g., *.py, test_*.ts)
• working_dir (optional): Directory to search in

Example:

{
  "action": "search_files",
  "parameters": {
    "pattern": "*.py",
    "working_dir": "/app/repos/agent-bot"
  }
}

search_code

Search for code patterns using grep.

Parameters:

• query (required): Text or regex pattern to search for
• file_types (optional): Filter by file extensions (default: *.py)

Example:

{
  "action": "search_code",
  "parameters": {
    "query": "async def process_",
    "file_types": ["*.py", "*.ts"]
  }
}

list_directory

List contents of a directory.

Parameters:

• path (optional): Path relative to working directory (default: current directory)

Example:

{
  "action": "list_directory",
  "parameters": {
    "path": "src/services"
  }
}

get_file_info

Get detailed information about a file.

Parameters:

• path (required): Path to the file

Example:

{
  "action": "get_file_info",
  "parameters": {
    "path": "src/main.py"
  }
}

find_references

Find all references to a symbol in the codebase.

Parameters:

• symbol (required): Symbol name to search for

Example:

{
  "action": "find_references",
  "parameters": {
    "symbol": "TaskWorker"
  }
}

get_project_structure

Get an overview of the project file structure.

Parameters:

• None

Example:

{
  "action": "get_project_structure",
  "parameters": {}
}

Integration with Knowledge Graph

For enhanced semantic search, use the knowledge-graph skill which provides:

• Semantic code search by function/class names
• Dependency graph navigation
• Call graph analysis
• Symbol reference tracking across the entire codebase

Workflow Example

Basic Discovery Flow

1. Get project structure:

   get_project_structure()
   

2. List specific directory:

   list_directory(path="src/services")
   

3. Search for relevant files:

   search_files(pattern="*service*.py")
   

4. Search for specific code:

   search_code(query="class.*Service")
   

5. Find symbol references:

   find_references(symbol="AuthService")
   

Combined with Knowledge Graph

For deeper analysis, combine with knowledge-graph skill:

1. Traditional search:

   search_code(query="async def authenticate")
   

2. Knowledge graph search:

   [Use knowledge-graph skill]
   search_codebase(query="authenticate", node_types=["function"])
   

3. Get dependencies:

   [Use knowledge-graph skill]
   find_dependencies(node_id="<result_id>")
   

Output Format

All actions return structured data:

{
  "success": true,
  "result": {
    "files": ["path/to/file1.py", "path/to/file2.py"],
    "count": 2
  }
}

Or on error:

{
  "success": false,
  "error": "Error message"
}