|
| 1 | +# code-context-graph |
| 2 | + |
| 3 | +Local code analysis tool that parses codebases via Tree-sitter into a knowledge graph. Supports 15 languages, 19 MCP tools, custom annotation search, and Apache AGE Cypher queries. |
| 4 | + |
| 5 | +## Features |
| 6 | + |
| 7 | +- **15 languages**: Go, Python, TypeScript, Java, Ruby, JavaScript, C, C++, Rust, C#, Kotlin, PHP, Swift, Scala, Lua, Bash |
| 8 | +- **19 MCP tools**: parse, search, impact analysis, flow tracing, dead code detection, Cypher queries, and more |
| 9 | +- **Custom annotations**: `@intent`, `@domainRule`, `@sideEffect`, `@mutates`, `@index` — search code by business context |
| 10 | +- **Apache AGE**: Cypher graph queries for path finding, blast-radius, pattern matching |
| 11 | +- **Multi-DB**: SQLite (local), PostgreSQL, MySQL |
| 12 | +- **Full-text search**: FTS5 (SQLite), tsvector+GIN (PostgreSQL), FULLTEXT (MySQL) |
| 13 | + |
| 14 | +## Quick Start |
| 15 | + |
| 16 | +### Build |
| 17 | + |
| 18 | +```bash |
| 19 | +CGO_ENABLED=1 go build -tags "fts5" -o ccg ./cmd/ccg/ |
| 20 | +``` |
| 21 | + |
| 22 | +### Parse your project |
| 23 | + |
| 24 | +```bash |
| 25 | +./ccg build . |
| 26 | +``` |
| 27 | + |
| 28 | +``` |
| 29 | +Build complete: 70 files, 749 nodes, 7387 edges |
| 30 | +``` |
| 31 | + |
| 32 | +### Search |
| 33 | + |
| 34 | +```bash |
| 35 | +# Keyword search (includes annotations) |
| 36 | +./ccg search "authentication" |
| 37 | + |
| 38 | +# Search by business context |
| 39 | +./ccg search "결제" # finds functions with @intent/@domainRule about payments |
| 40 | +./ccg search "dead code" # finds deadcode.Find via @intent annotation |
| 41 | +``` |
| 42 | + |
| 43 | +### Status |
| 44 | + |
| 45 | +```bash |
| 46 | +./ccg status |
| 47 | +``` |
| 48 | + |
| 49 | +``` |
| 50 | +Nodes: 747 |
| 51 | +Edges: 6780 |
| 52 | +Files: 70 |
| 53 | +
|
| 54 | +Node kinds: |
| 55 | + function: 238 |
| 56 | + test: 339 |
| 57 | + class: 83 |
| 58 | + type: 17 |
| 59 | + file: 70 |
| 60 | +
|
| 61 | +Edge kinds: |
| 62 | + calls: 5124 |
| 63 | + contains: 679 |
| 64 | + tested_by: 543 |
| 65 | + imports_from: 434 |
| 66 | +``` |
| 67 | + |
| 68 | +## Custom Annotations |
| 69 | + |
| 70 | +Add structured metadata to your code. These are indexed and searchable. |
| 71 | + |
| 72 | +### File-level |
| 73 | + |
| 74 | +```go |
| 75 | +// @index User authentication and session management service. |
| 76 | +package auth |
| 77 | +``` |
| 78 | + |
| 79 | +### Function-level |
| 80 | + |
| 81 | +```go |
| 82 | +// AuthenticateUser validates credentials and creates a session. |
| 83 | +// Called from login API handler. |
| 84 | +// |
| 85 | +// @param username user login ID |
| 86 | +// @param password plaintext password |
| 87 | +// @return JWT token on success |
| 88 | +// @intent verify user identity before granting system access |
| 89 | +// @domainRule lock account after 5 consecutive failed attempts |
| 90 | +// @sideEffect writes login attempt to audit_log table |
| 91 | +// @mutates user.FailedAttempts, user.LockedUntil |
| 92 | +// @requires user.IsActive == true |
| 93 | +// @ensures err == nil implies valid JWT with 24h expiry |
| 94 | +func AuthenticateUser(username, password string) (string, error) { |
| 95 | +``` |
| 96 | +
|
| 97 | +### Available Tags |
| 98 | +
|
| 99 | +| Tag | Purpose | Example | |
| 100 | +|-----|---------|---------| |
| 101 | +| `@index` | File/package description | `@index Payment processing service` | |
| 102 | +| `@intent` | Why this function exists | `@intent verify credentials before session creation` | |
| 103 | +| `@domainRule` | Business rule | `@domainRule lock account after 5 failures` | |
| 104 | +| `@sideEffect` | Side effects | `@sideEffect sends notification email` | |
| 105 | +| `@mutates` | State changes | `@mutates user.FailedAttempts, session.Token` | |
| 106 | +| `@requires` | Precondition | `@requires user.IsActive == true` | |
| 107 | +| `@ensures` | Postcondition | `@ensures session != nil` | |
| 108 | +| `@param` | Parameter description | `@param username the login ID` | |
| 109 | +| `@return` | Return description | `@return JWT token on success` | |
| 110 | +| `@see` | Related function | `@see SessionManager.Create` | |
| 111 | +
|
| 112 | +## Apache AGE (Graph Queries) |
| 113 | +
|
| 114 | +### Setup |
| 115 | +
|
| 116 | +```bash |
| 117 | +docker compose up age -d |
| 118 | +``` |
| 119 | +
|
| 120 | +### Build with graph sync |
| 121 | +
|
| 122 | +```bash |
| 123 | +AGE_DSN="host=127.0.0.1 port=5455 dbname=ccg user=ccg password=ccg sslmode=disable" \ |
| 124 | + ./ccg build --graph . |
| 125 | +``` |
| 126 | +
|
| 127 | +### Cypher queries |
| 128 | +
|
| 129 | +```bash |
| 130 | +# All function call relationships |
| 131 | +./ccg query "MATCH (a:Function)-[:CALLS]->(b:Function) RETURN a.name, b.name" --columns 2 |
| 132 | + |
| 133 | +# Blast-radius (3 hops) |
| 134 | +./ccg query "MATCH ({name: 'Login'})-[*1..3]-(n) RETURN DISTINCT n.qualified_name" |
| 135 | + |
| 136 | +# Call path between two functions |
| 137 | +./ccg query "MATCH path = (a {name: 'Handler'})-[:CALLS*]->(b {name: 'Save'}) RETURN path" |
| 138 | + |
| 139 | +# Dead code |
| 140 | +./ccg query "MATCH (n:Function) WHERE NOT ()-[:CALLS]->(n) RETURN n.qualified_name" |
| 141 | + |
| 142 | +# Most called functions |
| 143 | +./ccg query "MATCH ()-[:CALLS]->(n) RETURN n.name, count(*) AS c ORDER BY c DESC LIMIT 10" |
| 144 | +``` |
| 145 | +
|
| 146 | +### Graph Schema |
| 147 | +
|
| 148 | +**Vertices**: `Function`, `Class`, `Type`, `Test`, `File` |
| 149 | +
|
| 150 | +**Edges**: `CALLS`, `IMPORTS_FROM`, `INHERITS`, `IMPLEMENTS`, `CONTAINS`, `TESTED_BY`, `DEPENDS_ON`, `REFERENCES` |
| 151 | +
|
| 152 | +## Claude Code Integration |
| 153 | +
|
| 154 | +### MCP Server |
| 155 | +
|
| 156 | +Add `.mcp.json` to your project: |
| 157 | +
|
| 158 | +```json |
| 159 | +{ |
| 160 | + "mcpServers": { |
| 161 | + "ccg": { |
| 162 | + "command": "./ccg", |
| 163 | + "args": ["serve", "--db", "sqlite", "--dsn", "ccg.db"] |
| 164 | + } |
| 165 | + } |
| 166 | +} |
| 167 | +``` |
| 168 | +
|
| 169 | +Claude Code automatically connects and gets access to 19 tools including `execute_cypher` for direct Cypher queries. |
| 170 | +
|
| 171 | +### Skill |
| 172 | +
|
| 173 | +The `/ccg` skill provides: |
| 174 | +
|
| 175 | +``` |
| 176 | +/ccg build . — Build code graph |
| 177 | +/ccg status — Graph statistics |
| 178 | +/ccg search "query" — Full-text search |
| 179 | +/ccg query "MATCH ..." — Cypher query |
| 180 | +/ccg annotate internal/ — AI-generate annotations |
| 181 | +``` |
| 182 | +
|
| 183 | +### AI-Driven Annotation |
| 184 | +
|
| 185 | +Claude can auto-generate annotations for your codebase: |
| 186 | +
|
| 187 | +``` |
| 188 | +You: "이 프로젝트에 어노테이션 달아줘" |
| 189 | +Claude: reads code → generates @intent, @domainRule, @sideEffect, @mutates |
| 190 | + → writes annotations → rebuilds index |
| 191 | + → now searchable by business context |
| 192 | +``` |
| 193 | +
|
| 194 | +## Architecture |
| 195 | +
|
| 196 | +``` |
| 197 | +Source Code → Tree-sitter Parser → Nodes + Edges + Annotations |
| 198 | + ↓ |
| 199 | + SQLite / PostgreSQL (GORM) |
| 200 | + ↓ ↓ |
| 201 | + FTS Search Apache AGE Graph |
| 202 | + ↓ ↓ |
| 203 | + MCP Server (19 tools) |
| 204 | + ↓ |
| 205 | + Claude Code |
| 206 | +``` |
| 207 | +
|
| 208 | +## CLI Commands |
| 209 | +
|
| 210 | +| Command | Description | |
| 211 | +|---------|-------------| |
| 212 | +| `ccg build [dir]` | Parse and build code graph | |
| 213 | +| `ccg build --graph [dir]` | Build + sync to Apache AGE | |
| 214 | +| `ccg build --embed [dir]` | Build + vector embeddings | |
| 215 | +| `ccg update [dir]` | Incremental sync | |
| 216 | +| `ccg status` | Graph statistics | |
| 217 | +| `ccg search <query>` | Full-text search | |
| 218 | +| `ccg search --semantic <q>` | Semantic vector search | |
| 219 | +| `ccg query <cypher>` | Execute Cypher query | |
| 220 | +| `ccg serve` | Start MCP server (stdio) | |
| 221 | +
|
| 222 | +## MCP Tools (19) |
| 223 | +
|
| 224 | +| Tool | Description | |
| 225 | +|------|-------------| |
| 226 | +| `parse_project` | Parse source files | |
| 227 | +| `build_or_update_graph` | Full/incremental build with postprocessing | |
| 228 | +| `run_postprocess` | Run flows/communities/search rebuild | |
| 229 | +| `get_node` | Get node by qualified name | |
| 230 | +| `search` | Full-text search | |
| 231 | +| `query_graph` | Predefined graph queries (callers, callees, imports, etc.) | |
| 232 | +| `list_graph_stats` | Node/edge/file counts | |
| 233 | +| `get_impact_radius` | BFS blast-radius analysis | |
| 234 | +| `trace_flow` | Call-chain flow tracing | |
| 235 | +| `find_large_functions` | Functions exceeding line threshold | |
| 236 | +| `find_dead_code` | Unused code detection | |
| 237 | +| `detect_changes` | Git diff risk scoring | |
| 238 | +| `get_affected_flows` | Flows affected by changes | |
| 239 | +| `list_flows` | List all traced flows | |
| 240 | +| `list_communities` | List module communities | |
| 241 | +| `get_community` | Community details + coverage | |
| 242 | +| `get_architecture_overview` | Architecture summary with coupling | |
| 243 | +| `get_annotation` | Get annotation and doc tags | |
| 244 | +| `execute_cypher` | Execute arbitrary Cypher queries | |
| 245 | +
|
| 246 | +## Development |
| 247 | +
|
| 248 | +```bash |
| 249 | +# Run tests |
| 250 | +CGO_ENABLED=1 go test -tags "fts5" ./... -count=1 |
| 251 | + |
| 252 | +# Build |
| 253 | +CGO_ENABLED=1 go build -tags "fts5" -o ccg ./cmd/ccg/ |
| 254 | + |
| 255 | +# Docker (PostgreSQL + AGE + MySQL) |
| 256 | +docker compose up -d |
| 257 | +``` |
| 258 | +
|
| 259 | +## License |
| 260 | +
|
| 261 | +MIT |
0 commit comments