Crg runs on your own machine — the client starts it, so there is no endpoint to ping. 2 447 installs a week from pypi. Last commit 16 Oct 2026.
Token-efficient code review knowledge graph: semantic search and call-graph resolution.
We read the source, 15 h ago · rules ccca72095570
What this server is able to do. For an MCP server this is often the job itself — a terminal server runs commands because that is what it is for. Listed so you know what you are plugging in, not as an accusation.
result = subprocess.run(
Is this your server and something here is wrong? Tell us — corrections are free and do not require a plan.
We found places where it runs commands, builds paths or queries from values it is given. None of that is a flaw by itself — it becomes one when the code changes, and code changes quietly between releases. We re-read it on every one.
This server runs on your own machine — install it with the package manager and the client starts it for you. Package name taken from the official registry entry.
claude mcp add crg -- uvx better-code-review-graph
{
"mcpServers": {
"crg": {
"args": [
"better-code-review-graph"
],
"command": "uvx"
}
}
}
[mcp_servers.crg]
command = "uvx"
args = ["better-code-review-graph"]
{
"mcpServers": {
"crg": {
"args": [
"better-code-review-graph"
],
"command": "uvx"
}
}
}
{
"mcpServers": {
"crg": {
"args": [
"better-code-review-graph"
],
"command": "uvx"
}
}
}
This one needs environment variables set before it will start:
EMBEDDING_MODELS (Embedding models in selection order; the current runtime uses the first provider/model entry; empty = local ONNX), LOCAL_EMBEDDING_MODEL (Built-in fastretrieval model ID or local directory with fastretrieval-manifest.json), LOCAL_RERANK_MODEL (Fastretrieval TextCrossEncoder model ID; empty = reranking disabled), LOCAL_EMBEDDING_DIM (Required positive output dimension for an external local embedding ID without a manifest), LOCAL_EMBEDDING_MODEL_FILE (ONNX file path inside a manifest-backed local embedding directory), LOCAL_EMBEDDING_POOLING (Explicit pooling for an external local embedding ID without a manifest), LOCAL_EMBEDDING_NORMALIZE (Explicit L2 normalization for an external local embedding ID without a manifest), HULL_EMBED_API_KEY (API key for the [models.embed] cell transport (OpenRouter pre-wired default); host-injected, overrides the cell api_key), HULL_CHAT_API_KEY (API key for the [models.chat] cell transport (OpenRouter pre-wired default); host-injected, enables LLM summaries when the cell model is set), EMBEDDING_BACKEND (DEPRECATED (honored one release): embedding backend override 'local' or 'cloud'. Use EMBEDDING_MODELS instead.).
The author declared them in the registry entry; get the values from the project itself.
Token-efficient code review knowledge graph: semantic search and call-graph resolution.
Local-first memory for Claude Code and any MCP client: hybrid search + knowledge graph, $0/token.
AI-powered code indexer with semantic search and knowledge graphs
Graph-powered code intelligence with semantic search and knowledge graph for AI assistants
AST-based semantic code search; results ship with their call graph (calls + callers).
Local code search for agents: symbols, usages, call graph, AST patterns, grep and semantic search
Token-efficient MCP server for codebase search, NuGet decompilation, and deep PDF reading.
Persistent knowledge memory for AI agents. Hybrid search, code graph, pgvector.
Answers built from our own checks of this server.