Interactive Shell

Overview

sqry shell starts an interactive REPL that keeps the index loaded in memory. Queries execute in single-digit milliseconds because the snapshot is already parsed and the cache is warm. It is the fastest way to explore a codebase interactively.

Start the shell

cd /path/to/your/project
sqry shell

The shell loads the index once on startup, then presents a prompt:

sqry v4.10.13 — 42,318 nodes, 67,204 edges
sqry>

Running queries

Type any query at the prompt — the same syntax as sqry query:

sqry> kind:function AND async:true
  src/api/routes.rs:42     handle_request
  src/auth/middleware.rs:18 validate_token
  ... (38 more results)

sqry> callers:validate_token
  src/api/routes.rs:45     handle_request
  src/tests/auth_test.rs:12 test_auth_flow

sqry> callers:handle_request --limit 5
  src/main.rs:28            main
  src/tests/api_test.rs:55  test_routes

All flags work the same as in the non-interactive sqry query command: --limit, --json, --kind, --lang, etc.

Graph commands

Run graph commands without the sqry graph prefix:

sqry> trace-path main database
sqry> direct-callers authenticate
sqry> cycles --type imports
sqry> unused --scope public
sqry> stats

History

The shell maintains query history across sessions. Use arrow keys to navigate previous queries, or search with Ctrl+R:

sqry> # Press Up/Down to cycle through history
sqry> # Press Ctrl+R to search history

Exiting

Type exit, quit, or press Ctrl+D to exit the shell.

When to use the shell

The shell is best for exploratory sessions — when you’re tracing through unfamiliar code, debugging a call chain, or iteratively narrowing down a set of symbols. Each query runs against the already-loaded index, so the 4ms warm-cache latency applies to every command after the first.

For scripted or CI use cases, use sqry batch instead.