explore

name	explore
description	Explore codebase with code intelligence and search across code and documents. Use when searching by concept, finding where functionality lives, or investigating unfamiliar code.

/explore

Find relevant code and documents. Use code intelligence first, then search for broader discovery.

When to Use

• Exploring unfamiliar code
• Searching by concept rather than exact name
• Finding "where does X happen?"
• Investigating functionality across code and docs

Steps

1. Try code intelligence first (symbol search):

   git-kb code symbols "<query>"
   

2. For matched symbols, explore structure:

   git-kb code callers "<matched-symbol>"
   git-kb code callees "<matched-symbol>"
   

3. Search KB documents for related context:

   git-kb search "<keywords>"
   

4. Use semantic search for broader discovery:

   git-kb ai semantic "<query>"
   

If MCP tools are available, prefer kb_symbols, kb_callers, kb_callees, kb_search, and kb_semantic for structured JSON output.

Output Format

## Exploring: "<query>"

### Code Matches

1. **src/services/auth.ts::login** (function) - Score: 0.89

export async function login(credentials: Credentials): Promise

Handles user authentication and token generation.

2. **src/middleware/auth.ts::validate** (function) - Score: 0.82

export function validate(token: Token): boolean

Validates JWT tokens.

### Document Matches

1. **specs/auth-spec** (spec) - Score: 0.85
"Authentication System Specification"

2. **tasks/my-task** (task) - Score: 0.78
"Auth Service Refactoring"

### Suggested Next Steps

- `/understand src/services/auth.ts` - Deep dive into auth module
- `/refactor-safety login` - Check impact before changes
- `git-kb show specs/auth-spec` - Read the auth spec

Scope Options

When using semantic search:

• scope: "all" — Search both code and documents (default)
• scope: "code" — Search only code symbols
• scope: "documents" — Search only KB documents

Prerequisites

For code intelligence, the index must be generated:

git-kb code index

For semantic search, embeddings must be generated:

git-kb ai embed

Document search uses FTS and does not require embeddings.