Commands to manage the CodeGraphContext engine, contexts, and global settings.

Self-diagnostic: validates the installation and default graph backend (FalkorDB Lite, FalkorDB Remote, KuzuDB, or Neo4j), Python version, and core dependencies.

Usage:

cgc doctor

Interactive wizard for AI client integration.

What it does:

1. Detects installed clients (Cursor, VS Code, Claude, and others).
2. Writes MCP config fragments (for example mcp.json).
3. Prepares environment variables for DEFAULT_DATABASE and backend credentials.

Usage:

cgc mcp setup

Wizard for the Neo4j server backend (Docker, local install, or Aura-style remote).

Usage:

cgc neo4j setup

Inspect and change settings without editing .env by hand. Persistent values are stored under ~/.codegraphcontext/.env unless a project-local .codegraphcontext/.env overrides them.

• cgc config show — Print merged configuration.
• cgc config set <key> <value> — Set a key (for example DEFAULT_DATABASE, SCIP_INDEXER, INDEX_SOURCE).
• cgc config db <backend> — Shortcut for cgc config set DEFAULT_DATABASE <backend>.
• cgc config reset — Restore defaults (with confirmation).

DEFAULT_DATABASE quick switch — valid values: falkordb, falkordb-remote, kuzudb, neo4j.

cgc config set DEFAULT_DATABASE falkordb-remote
cgc config db falkordb
cgc config db falkordb-remote
cgc config db kuzudb
cgc config db neo4j

Named and per-repo contexts isolate databases and ignore rules. Configuration lives in ~/.codegraphcontext/config.yaml plus optional per-repo .codegraphcontext/.

• cgc context list — Show mode (global / per-repo / named), default named context, and all registered contexts with database type and paths.
• cgc context create <name> — Create a named context. Optional --database / -d (falkordb, kuzudb, neo4j; defaults to DEFAULT_DATABASE). Optional --db-path for a custom on-disk location.
• cgc context delete <name> — Remove a context from the registry (database files remain on disk unless you delete them manually).
• cgc context mode <global|per-repo|named> — Set how the CLI resolves which graph to use.
• cgc context default <name> — Default named context when mode is named and no --context flag is passed.

Examples:

cgc context list
cgc context create MyMonolith --database kuzudb
cgc context mode per-repo
cgc context default MyMonolith
cgc context delete OldProject

Index or query with a specific named context using --context / -c on commands that support it (for example cgc index . --context MyMonolith).

• Graph queries: use cgc query for read-only Cypher. The hidden alias cgc cypher remains for backward compatibility but prints a deprecation notice—prefer cgc query.
• Structured search: use the cgc find command group (cgc find name, cgc find pattern, and other subcommands), not a legacy cgc search command.

cgc query "MATCH (f:Function) RETURN f.name LIMIT 10"
cgc find name MyClass

• Configuration reference — DEFAULT_DATABASE, SCIP flags, INDEX_SOURCE, and backend connection variables.
• MCP tool overview — Tools exposed to IDEs, including discover_codegraph_contexts and switch_context.