opencode-codebase-index 0.3.0 → 0.4.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "opencode-codebase-index",
-  "version": "0.3.0",
+  "version": "0.4.0",
   "description": "Semantic codebase indexing and search for OpenCode - find code by meaning, not just keywords",
   "type": "module",
   "main": "dist/index.js",
@@ -38,6 +38,7 @@
     "dev": "tsup --watch",
     "test": "vitest",
     "test:run": "vitest run",
+    "test:coverage": "vitest run --coverage",
     "lint": "eslint src/",
     "typecheck": "tsc --noEmit",
     "prepublishOnly": "npm run build"
@@ -70,6 +71,7 @@
     "@napi-rs/cli": "^3.5.1",
     "@opencode-ai/plugin": "^1.1.21",
     "@types/node": "^25.0.8",
+    "@vitest/coverage-v8": "^4.0.17",
     "eslint": "^9.0.0",
     "tsup": "^8.1.0",
     "typescript": "^5.5.0",

package/skill/SKILL.md

@@ -1,191 +1,66 @@
-# Codebase Search Skill
+---
+name: codebase-search
+description: Semantic code search by meaning. Use codebase_peek to find WHERE code is (saves tokens), codebase_search to see actual code. For exact identifiers, use grep instead.
+---
 
-Semantic search for finding code by meaning, not just keywords.
+# Codebase Search Skill
 
-## When to Use Semantic Search vs Grep
+## When to Use What
 
-| Scenario | Use | Why |
-|----------|-----|-----|
-| Don't know function/class names | `codebase_search` | Natural language → code |
-| Concept discovery ("what handles auth?") | `codebase_search` | Finds related code by meaning |
+| Scenario | Tool | Why |
+|----------|------|-----|
+| Just need file locations | `codebase_peek` | Metadata only, saves ~90% tokens |
+| Need to see actual code | `codebase_search` | Returns full code content |
+| Find duplicates/patterns | `find_similar` | Given code snippet → similar code |
+| Don't know function/class names | `codebase_peek` or `codebase_search` | Natural language → code |
 | Know exact identifier names | `grep` | Faster, more precise |
 | Need ALL occurrences | `grep` | Semantic returns top N only |
-| Unfamiliar codebase exploration | `codebase_search` | Broader conceptual matches |
-
-## Hybrid Strategy (Recommended)
-
-1. **Discover with semantic**: `codebase_search("authentication flow")` → get candidate files
-2. **Drill down with grep**: `grep "validateToken"` in those files for exact matches
-3. **Use LSP**: For definitions and references once you have the identifier
-
-## Available Tools
-
-### `codebase_search`
-Find code by describing what it does in natural language.
 
-**Best for:**
-- "find the user authentication logic"
-- "code that handles database connections"  
-- "error handling middleware for HTTP requests"
-- "where is the payment processing"
+## Recommended Workflow
 
-**Parameters:**
-- `query` (required): Natural language description
-- `limit` (optional): Maximum results (default: 10)
-- `chunkType` (optional): Filter by code type
-- `directory` (optional): Filter by directory path
-- `fileType` (optional): Filter by file extension
-- `contextLines` (optional): Extra lines before/after each match
+1. **Locate with peek**: `codebase_peek("authentication flow")` → get file locations
+2. **Read what matters**: `Read` the specific files you need
+3. **Drill down with grep**: `grep "validateToken"` for exact matches
 
-**Returns:** Focused list of 5-10 most relevant files/chunks with scores.
-
-### `index_codebase`
-Create or update the semantic index. Required before first search.
-
-**Parameters:**
-- `force` (optional): Reindex from scratch (default: false)
-- `estimateOnly` (optional): Show cost estimate without indexing
-- `verbose` (optional): Show skipped files and parse failures
-
-**Note:** Incremental indexing is fast (~50ms) when files haven't changed.
-
-### `index_status`
-Check if the codebase is indexed and ready for search.
+## Tools
 
-### `index_health_check`
-Remove stale entries from deleted files and orphaned embeddings.
-
-## Search Filters
-
-### Filter by Chunk Type (`chunkType`)
-
-Narrow results to specific code constructs:
-
-| Value | Finds |
-|-------|-------|
-| `function` | Functions, arrow functions |
-| `class` | Class definitions |
-| `method` | Class methods |
-| `interface` | TypeScript interfaces |
-| `type` | Type aliases |
-| `enum` | Enumerations |
-| `struct` | Rust/Go structs |
-| `impl` | Rust impl blocks |
-| `trait` | Rust traits |
-| `module` | Module definitions |
-
-**Examples:**
-```
-codebase_search(query="validation logic", chunkType="function")
-codebase_search(query="data models", chunkType="interface")
-codebase_search(query="user entity", chunkType="class")
-```
-
-### Filter by Directory (`directory`)
-
-Scope search to specific paths:
+### `codebase_peek`
+Find WHERE code is. Returns metadata only (file, line, name, type).
 
 ```
-codebase_search(query="API routes", directory="src/api")
-codebase_search(query="test helpers", directory="tests")
-codebase_search(query="database queries", directory="src/db")
+codebase_peek(query="validation logic", chunkType="function", directory="src/utils")
 ```
 
-### Filter by File Type (`fileType`)
-
-Limit to specific languages:
+### `codebase_search`
+Find code with full content. Use when you need to see implementation.
 
 ```
-codebase_search(query="config parsing", fileType="ts")
-codebase_search(query="build scripts", fileType="py")
-codebase_search(query="data structures", fileType="rs")
+codebase_search(query="error handling middleware", fileType="ts", contextLines=2)
 ```
 
-### Combining Filters
-
-Filters can be combined for precise results:
+### `find_similar`
+Find code similar to a given snippet. Use for duplicate detection, pattern discovery, refactoring.
 
 ```
-codebase_search(
-  query="validation",
-  chunkType="function",
-  directory="src/utils",
-  fileType="ts"
-)
+find_similar(code="function validate(input) { return input.length > 0; }", excludeFile="src/current.ts")
 ```
 
-## Hybrid Weight Tuning
-
-The `hybridWeight` config option (0.0-1.0) balances semantic vs keyword search:
-
-| Value | Behavior | Best For |
-|-------|----------|----------|
-| `0.0` | Pure semantic | Conceptual queries, unfamiliar code |
-| `0.5` | Balanced (default) | General use |
-| `1.0` | Pure keyword (BM25) | When you know specific terms |
-
-Configure in `.opencode/codebase-index.json`:
-```json
-{
-  "search": {
-    "hybridWeight": 0.3
-  }
-}
-```
+### `index_codebase`
+Create/update index. Required before first search. Incremental (~50ms when unchanged).
 
-**When to adjust:**
-- Lower (0.2-0.4): Exploratory queries, finding related code
-- Higher (0.6-0.8): When query contains specific identifiers
+### `index_status`
+Check if indexed and ready.
 
-## Query Writing Tips
+## Query Tips
 
 **Describe behavior, not syntax:**
-```
-Good: "function that hashes passwords securely"
-Bad:  "hashPassword" (use grep for exact names)
-
-Good: "middleware that checks JWT tokens"
-Bad:  "jwt" (too vague, use grep for keyword)
-
-Good: "error handling for unauthorized requests"  
-Bad:  "401" (literal keyword, use grep)
-```
-
-## Workflow Examples
-
-### Exploring unfamiliar codebase
-```
-1. index_status → check if indexed
-2. index_codebase → if needed
-3. codebase_search("how does authentication work") → get overview
-4. codebase_search("session management") → drill into specifics
-5. grep "SessionStore" → once you know the class name
-```
-
-### Finding all functions in a module
-```
-1. codebase_search(query="utility helpers", directory="src/utils", chunkType="function")
-2. Review results to understand available utilities
-3. grep specific function name for all usages
-```
-
-### Finding implementation for a feature
-```
-1. codebase_search("image upload and processing") → find relevant files
-2. Read top results to understand structure
-3. grep "uploadImage" → find exact function
-4. LSP goto_definition → navigate to implementation
-```
+- Good: `"function that hashes passwords securely"`
+- Bad: `"hashPassword"` (use grep for exact names)
 
-### Debugging unknown code path
-```
-1. codebase_search("error handling for payment failures") → find error handlers
-2. codebase_search("retry logic for API calls") → find retry mechanisms
-3. grep "PaymentError" → find specific error class
-```
+## Filters
 
-### Finding TypeScript interfaces
-```
-1. codebase_search(query="user data", chunkType="interface") → find User interfaces
-2. codebase_search(query="API response", chunkType="type") → find response types
-```
+| Filter | Example |
+|--------|---------|
+| `chunkType` | `function`, `class`, `interface`, `type`, `method` |
+| `directory` | `"src/api"`, `"tests"` |
+| `fileType` | `"ts"`, `"py"`, `"rs"` |