Initialize

cd your-project
dora init
dora index

dora init creates .dora/config.json. dora index runs the SCIP indexer and converts the output to SQLite.

Usage

Exploring a codebase

dora map                          # file count, symbol count, packages
dora ls src/                      # files in a directory with symbol/dep counts
dora status                       # index health and grammar availability

Finding code

dora symbol AuthService           # find symbols by name
dora file src/auth/service.ts     # symbols, dependencies, dependents for a file
dora refs validateToken           # all references to a symbol across the codebase

Understanding dependencies

dora deps src/auth/service.ts --depth 2    # what this file imports
dora rdeps src/auth/service.ts --depth 2   # what imports this file
dora adventure src/a.ts src/b.ts           # shortest path between two files

Tree-sitter analysis (TypeScript / JavaScript)

These commands parse source directly without needing an index:

dora fn src/auth/service.ts       # functions with complexity, params, return type, LOC
dora class src/auth/service.ts    # classes with methods, implements, decorators
dora smells src/auth/service.ts   # high complexity, long functions, too many params, TODOs

Install a grammar to enable these:

bun add -g tree-sitter-typescript

Architecture

dora cycles                        # bidirectional dependencies (A imports B, B imports A)
dora coupling --threshold 5        # file pairs with high symbol sharing
dora complexity --sort complexity  # files ranked by change risk
dora treasure                      # most imported files
dora lost                          # symbols with zero references

Documentation

dora docs                          # list indexed markdown/text files
dora docs search "authentication"  # full-text search across docs
dora docs show docs/api.md         # which symbols and files a doc references

Custom queries

dora schema                        # database schema
dora cookbook show quickstart      # walkthrough with real SQL examples
dora cookbook show methods         # query patterns for finding methods
dora query "SELECT path, symbol_count FROM files ORDER BY symbol_count DESC LIMIT 10"

MCP server

dora mcp                           # start MCP server (stdio)

# Claude Code
claude mcp add --transport stdio dora -- dora mcp

Output format

All commands output TOON by default — a compact JSON encoding optimized for LLM token usage. Pass --json for standard JSON.

dora status --json

AI agent integration

For agent-specific setup (hooks, skills, AGENTS.md snippets) for Claude Code, OpenCode, Cursor, and Windsurf:

dora cookbook show agent-setup

Or see AGENTS.README.md.

How it works

dora has two layers:

SCIP layer — runs your configured indexer (e.g. scip-typescript) to produce a .scip protobuf file, then parses it and loads it into SQLite. This gives you symbol definitions, references, and file-to-file dependencies derived from actual import resolution.

Tree-sitter layer — parses source files directly using WebAssembly grammars. This runs on-demand per file and extracts things SCIP doesn't cover: function signatures, cyclomatic complexity, class hierarchy details, and code smells.

The SQLite schema uses denormalized counts (symbol_count, dependency_count, dependent_count, reference_count) so most queries are index lookups rather than aggregations.

.dora/
├── config.json      # indexer command, ignore patterns, grammar paths
├── index.scip       # raw SCIP protobuf output
└── dora.db          # SQLite database (the thing dora actually queries)

Contributing

See CONTRIBUTING.md.

License

MIT