Quickstart

From zero to a first successful query in about five minutes.

No Rust toolchain, no compiler — a prebuilt binary. Full details, including downloading it by hand: Installation.

# 1. Install
curl -fsSL https://raw.githubusercontent.com/alexeyban/EKOS/main/install.sh | sh

# 2. Compile your project
cd /path/to/my-project
ekos init --detect                # writes an ekos.toml matching this repository
ekos build && ekos recover && ekos resolve && ekos compile && ekos commit
ekos coverage                     # confirms every input kind produced objects

# 3. Query it
ekos query find "customer"

ekos init --detect beats a bare ekos init: it detects the SQL dialect your schema is written in and excludes third-party and generated directories that would otherwise be compiled as if they were your own code. A wrong dialect makes the whole-file DDL parse fail, and an entire schema disappears behind one diagnostic.

ekos coverage is the other half of that: it reports which input kinds actually compiled into objects. A kind with files present and zero objects is a named finding, not silence.

If ekos resolve stops on identity conflicts, review them, or pass --force to continue and print them as diagnostics.

Connect Claude Code

claude mcp add ekos -- ekos --config /path/to/my-project/ekos.toml mcp serve --workspace /path/to/my-project

Then ask:

Claude Code will call tools such as ekos_search, ekos_query and ekos_state and cite the evidence it gets back.

Where to go from here