better-code-review-graph

Knowledge graph for token-efficient code reviews with Tree-sitter parsing, dual-mode embedding (ONNX + LiteLLM), and blast-radius analysis via MCP tools.

GitHub

Documentation

Better Code Review Graph

Renamed (2026-09-13): repo is now crg — CLI-first (crg command). PyPI package stays better-code-review-graph; MCP server is a secondary surface.

mcp-name: io.github.n24q02m/better-code-review-graph

Knowledge graph for token-efficient code reviews -- semantic search and call-graph resolution across your codebase.

Sister projects from n24q02m (click to expand)

Project	Tagline	Tag
agent-chat-plugin	Peer AI agents chat in a shared folder — no human relay, no orchestrator, wor...	Tooling
better-code-review-graph	Knowledge graph for token-efficient code reviews -- semantic search and call-...	MCP
better-drive	2-way Google Drive sync with .driveignore filter — rclone engine, Windows tray	Tooling
better-email-mcp	IMAP/SMTP email for AI agents -- read, send, organize folders, and manage att...	MCP
better-godot-mcp	Composite MCP server for Godot Engine -- 17 composite tools for AI-assisted g...	MCP
better-notion-mcp	Markdown-first Notion for AI agents -- pages, databases, blocks, and comments...	MCP
better-semantic-release	Drop-in python-semantic-release fork with built-in release-safety guards (orp...	Tooling
better-telegram-mcp	Telegram for AI agents -- messages, chats, media, and contacts across both bo...	MCP
better-workspace-mcp	Google Workspace MCP server (Docs/Drive/Calendar/Gmail/Sheets/Slides/Tasks/Ch...	MCP
claude-plugins	Claude Code plugin marketplace for the n24q02m MCP servers -- install web sea...	Marketplace
imagine-mcp	Image and video understanding + generation for AI agents -- across Gemini, Op...	MCP
jules-task-archiver	Chrome Extension for bulk operations on Jules tasks via batchexecute API -- a...	Tooling
mcp-core	Shared foundation for building MCP servers -- Streamable HTTP transport, OAut...	MCP
mnemo-mcp	Persistent AI memory with hybrid search and embedded sync. Open, free, unlimi...	MCP
fastretrieval	Fast multi-model retrieval runtime for ONNX and GGUF embeddings, reranking, and model contracts	Library
skret	Secrets without the server.	CLI
tacet	A self-distilling neuro-symbolic cascade that amortises LLM cost across knowl...	Tooling
web-core	Shared web infrastructure package for search, scraping, HTTP security, and st...	Library
wet-mcp	Open-source MCP server for AI agents: web search, content extraction, and lib...	MCP

An MCP server that parses your codebase with Tree-sitter, builds a structural graph of functions/classes/imports, and gives Claude (or any MCP client) precise context so it reads only what matters instead of the whole tree. Semantic search runs through the local ONNX model registry from fastretrieval by default (zero config, no API key), with an optional cloud embedding chain. Fork of code-review-graph with fixed multi-word search, qualified call resolution, dual-mode embeddings, output pagination, and production CI/CD.

v2.0 migration (BREAKING)

v2.0 adds temporal columns (valid_from_sha / valid_to_sha on every node + edge) and an opt-in security scanner. The schema migration is auto-applied on first GraphStore open, and a backup of the pre-2.0 DB is saved to <graph_db>.pre-2.0.bak so you can roll back. See BREAKING_CHANGES.md for the full schema-change list, behavior changes, environment requirements, and the downgrade procedure (CRG_DOWNGRADE_TO_1_X=1 uv run better-code-review-graph).

v2.0 migration (BREAKING)
Install
Smithery
Configuration
Tools
CLI
Features
Comparison
Security
Build from source
Trust model
Migration & changelog
Documentation
License

Install

For OMP and other local coding harnesses, the primary surface is the package CLI plus the bundled skills/ workflows. The skills invoke the CLI directly and do not require an MCP server mapping.

# Run without a persistent install (short `crg` script; PyPI package name stays
# better-code-review-graph, so `uvx` needs the explicit --from form)
uvx --python 3.13 --from better-code-review-graph crg graph build --full-rebuild \
  --repo-root /path/to/repo
uvx --python 3.13 --from better-code-review-graph crg graph stats \
  --repo-root /path/to/repo

# Or install the console scripts (installs both `crg` and the legacy long name)
pip install better-code-review-graph
crg query search --search-query "authentication" \
  --repo-root /path/to/repo

The optional Semgrep engine for deeper security scans is a separate extra:

pip install 'better-code-review-graph[security]'

MCP stdio remains a secondary protocol adapter for clients that require it:

{
  "mcpServers": {
    "better-code-review-graph": {
      "command": "uvx",
      "args": ["--python", "3.13", "better-code-review-graph"],
      "env": { "MCP_TRANSPORT": "stdio" }
    }
  }
}

Install matrix (stdio unless noted; the CLI-first usage above stays the primary surface):

Client	Install
Claude Code (plugin)	`/plugin marketplace add n24q02m/claude-plugins` then `/plugin install better-code-review-graph@n24q02m-plugins`
Claude Code (stdio)	`claude mcp add better-code-review-graph -- uvx --python 3.13 better-code-review-graph`
Codex	register stdio command `uvx --python 3.13 better-code-review-graph` under `mcp_servers` in `~/.codex/config.toml`
Gemini CLI	add the `mcpServers` JSON above to `~/.gemini/settings.json`
Cursor / Windsurf	add the `mcpServers` JSON above via the client's MCP settings (`mcp.json`)
Any client (HTTP self-host)	point the client at `https://<your-host>/mcp` (`MCP_TRANSPORT=http`) -- self-host only, no hosted endpoint

Install with an AI agent -- paste this to your AI coding agent:

Install MCP server better-code-review-graph following the steps at https://raw.githubusercontent.com/n24q02m/claude-plugins/main/plugins/better-code-review-graph/setup-with-agent.md

Full CLI usage is in CLI. Optional per-client MCP setup is at mcp.n24q02m.com/servers/better-code-review-graph/setup/.

Local-first boundary

CRG is local-first for coding workflows:

CLI and bundled Skills are the primary surfaces for graph build/query, impact analysis, review context, security scans, and repository onboarding.
MCP stdio is the secondary protocol adapter over the same local domain services; it does not maintain a separate graph implementation.
Graph state stays in <repo>/.better-code-review-graph/graph.db unless an explicit multi-user/self-host configuration selects another data directory.
PyPI, CI, security scanning, GitHub releases, and eligible stable MCP Registry publication remain active. Historical public OCI tags are retained, but new public Docker Hub/GHCR images are no longer published.
CRG has no hosted Cloudflare runtime in the target topology.

Smithery

The repo ships a smithery.yaml so the server can be built and run through Smithery. It deploys over stdio and needs no startup configuration -- the config schema is empty, and any optional cloud embedding/summary keys are supplied at runtime through the server's own config flow (see Configuration below). The launch command is the same uvx invocation as a local install:

startCommand:
  type: stdio
  commandFunction: |-
    (config) => ({ command: 'uvx', args: ['--python', '3.13', 'better-code-review-graph'] })

Configuration

Everything works out of the box with zero configuration -- semantic search uses the local ONNX registry from fastretrieval (Qwen3-Embedding-0.6B is the current built-in reference entry, ~570 MB downloaded on first graph embed). This reference entry is not a Qwen-only boundary: any built-in registry ID or valid non-Qwen artifact manifest follows the same resolver. All environment variables below are optional and only needed for cloud embeddings, LLM summaries, or an explicit BYO local artifact.

Model selection

Embeddings select the first provider/model entry in EMBEDDING_MODELS; later entries are retained as configuration but are not runtime fallbacks. Summaries select the first SUMMARY_MODELS entry too, without runtime fallback. Providers are inferred from model prefixes and use the matching <PROVIDER>_API_KEY.

Variable	Purpose	Empty (default)
`EMBEDDING_MODELS`	Cloud embedding selection; the first entry is active	Local fastretrieval registry
`SUMMARY_MODELS`	Completion model selection for `graph(action="summarize")`	Summaries disabled

Cohere embed-v4.0 requests and stores 1024 dimensions; other backends retain 768-dimensional storage. CRG never slices, pads, or silently accepts a different provider width. The embedding row's model and byte width must match before reuse. Run graph(action="embed") after changing models or upgrading an old 768-wide Cohere index. Searches reject incompatible widths before a provider call; graph nodes are retained and re-embedding replaces only stale vectors.

Provider API keys

Cloud models need the provider key for the selected model prefix. Keys alone never select models: an empty embedding chain stays local, and an empty summary chain stays disabled. A configured cloud error does not fall back to local or another provider. Summarizers require a chat-completion model.

Model prefix	API key env var	Get a key
`jina_ai/`	`JINA_AI_API_KEY`	https://jina.ai/api-key
`gemini/`	`GEMINI_API_KEY` (or `GOOGLE_API_KEY`)	https://aistudio.google.com/apikey
`openai/` (or bare `text-embedding-*`)	`OPENAI_API_KEY`	https://platform.openai.com/api-keys
`cohere/`	`COHERE_API_KEY`	https://dashboard.cohere.com/api-keys
`openrouter/`	`OPENROUTER_API_KEY`	https://openrouter.ai/settings/keys
`vertex_express/`	`GOOGLE_VERTEX_EXPRESS_API_KEY`	https://cloud.google.com/vertex-ai/generative-ai/docs/start/express-mode/overview

Any other litellm provider works via its standard <PROVIDER>_API_KEY.

Advanced

Variable	Purpose
`EMBEDDING_API_BASE`	Provider-compatible endpoint for cloud embedding, including CF AI Gateway (SSRF-guarded)
`LLM_API_BASE`	Provider-compatible base URL for the summarizer, including CF AI Gateway (SSRF-guarded)
`DISABLE_LOCAL_EMBED`	Skip the local ONNX download; embedding is unavailable unless a cloud chain is configured
`LOCAL_EMBEDDING_MODEL`	Built-in fastretrieval model ID, or a local directory containing `fastretrieval-manifest.json`
`LOCAL_RERANK_MODEL`	Fastretrieval `TextCrossEncoder` model ID for bounded semantic reranking
`LOCAL_EMBEDDING_DIM`	Required dimension for an external model ID without a manifest
`LOCAL_EMBEDDING_MODEL_FILE`	ONNX file path inside a manifest-backed artifact directory
`LOCAL_EMBEDDING_POOLING`	Explicit pooling for an external ID without a manifest: `CLS`, `MEAN`, `LAST_TOKEN`, or `DISABLED`
`LOCAL_EMBEDDING_NORMALIZE`	Explicit L2 normalization for an external ID without a manifest
`CRG_DATA_DIR`	Override the per-user data directory (default `~/.crg`) used for per-user graphs and credentials in HTTP multi-user mode
`EMBEDDING_BACKEND` / `EMBEDDING_MODEL` / `SUMMARY_MODEL`	Deprecated singular vars, honored one release with a warning -- migrate to the `*_MODELS` chains

When LOCAL_RERANK_MODEL is configured, semantic vector search retrieves a bounded candidate pool of min(max(limit * 4, limit), 100) rows, applies the existing kind, repo, and live-row filters, then reranks that pool and returns at most limit rows. The response uses search_mode="semantic_reranked" and adds rerank_score while preserving similarity_score. Blank keeps the existing limit * 2 vector path and search_mode="semantic". Configured reranker failures return an explicit error; CRG does not silently fall back to vector or keyword results. Keyword searches, including as_of snapshots, do not invoke the reranker.

Example -- cloud embeddings + summaries

{
  "mcpServers": {
    "better-code-review-graph": {
      "command": "uvx",
      "args": ["--python", "3.13", "better-code-review-graph"],
      "env": {
        "MCP_TRANSPORT": "stdio",
        "EMBEDDING_MODELS": "cohere/embed-v4.0",
        "SUMMARY_MODELS": "openrouter/minimax/minimax-m3:free",
        "EMBEDDING_API_BASE": "https://gateway.ai.cloudflare.com/v1/<account>/<gateway>/cohere/v2/embed",
        "LLM_API_BASE": "https://gateway.ai.cloudflare.com/v1/<account>/<gateway>/openrouter/v1",
        "COHERE_API_KEY": "<cohere-key>",
        "OPENROUTER_API_KEY": "<openrouter-key>"
      }
    }
  }
}

Cohere embedding is paid. Authorize a bounded budget before a live index/query; the Minimax-free completion choice does not make embeddings free. This example does not add a process-wide model override: in remote mode each authenticated subject's relay record supplies its models, endpoints and keys. Missing subject credentials fail closed rather than inheriting the server environment.

CRG currently has no cloud rerank call: LOCAL_RERANK_MODEL is its only reranking path. Setting RERANK_MODELS or RERANK_API_BASE does not enable one.

You can also configure cloud keys interactively in HTTP mode via the relay setup form (config(action="setup_start") returns the browser URL). See the modes overview and multi-user setup.

Workspace username (HTTP setup form)

The relay setup form has an optional workspace username field. Entering the same username always lands you in the same per-sub bucket, so your keys and graph stay reachable across a re-authorization and across devices, instead of being tied to the one-off subject minted for each /authorize round-trip. Leaving it blank keeps the previous per-authorize behaviour.

Trust boundary: when the form is gated by a shared MCP_RELAY_PASSWORD, the username is a partition key, not a secret -- anyone who knows that password can type any username and reach that bucket. That is fine for a trusted group; an untrusted multi-tenant deployment needs a per-user secret or delegated OAuth instead.

One-time migration: existing users must re-enter their credentials once after this change. Nothing is deleted; credentials stored under the old random subject are simply no longer addressed.

Tools

Seven tools, each grouping related actions to keep the tool surface small.

`graph` -- Graph lifecycle

Action	Description
`build`	Full or incremental graph build. Set `full_rebuild=true` to re-parse all files; pass `roots` to federate extra repo directories into one graph.
`update`	Alias for `build` with `full_rebuild=false` (incremental).
`stats`	Graph size, languages, node/edge breakdown, embedding count.
`embed`	Compute vector embeddings for semantic search. Dual-mode: local ONNX or cloud chain.
`export`	Export the graph as `graphml` / `json-ld` / `dot` / `cypher`. Inline or to `output_path`.
`summarize`	LLM-generated one-paragraph docstrings for `Function` nodes (via the first explicit `SUMMARY_MODELS` entry; no-op when no model is selected). Calls bounded by `max_nodes`.

`query` -- Graph queries

Action	Description
`query`	Predefined patterns: `callers_of`, `callees_of`, `imports_of`, `importers_of`, `children_of`, `tests_for`, `inheritors_of`, `file_summary`.
`search`	Search code entities by name/keyword or semantic similarity.
`impact`	Blast radius of changed files. Auto-detects from git diff. Paginated with `max_results`.
`large_functions`	Find functions/classes exceeding a line-count threshold.
`spot_check`	Random callsite snippets from the last `callers_of`/`callees_of`/`inheritors_of`/`importers_of` result.
`renamed_in_diff`	Symbols whose callsite line shifted versus a base ref.
`diff`	Nodes added/removed/modified between two commit SHAs (`from_sha`, `to_sha`).

Most read actions accept as_of=<sha> for temporal (point-in-time) snapshots and repo=<repo_id> to scope a federated multi-repo graph.

`review` -- Code review context

Actions: context (default) | delta

Token-optimized review context with structural summary, impacted nodes, source snippets, and review guidance. context auto-detects changed files from the git diff; delta (with from_sha/to_sha, optional show_line_shifts) surfaces refactor moves between two commits.

`config` -- Server configuration and credential setup

Action	Description
`status`	Server info: version, graph path, node/edge counts, embedding backend, embeddings count.
`set`	Update a runtime setting (`key=log_level`).
`cache_clear`	Remove all computed embeddings.
`setup_status`	Show current credential state, providers configured, and setup URL.
`setup_start`	Start relay setup to configure API keys via browser (HTTP mode).
`setup_skip`	Set local mode (skip relay permanently, use ONNX only).
`setup_reset`	Clear credentials and reset state.
`setup_complete`	Re-resolve credentials from environment variables.

`security` -- Security scanning

Actions: scan | report | suppress | rule_list

Action	Description
`scan`	Run a security scan (`engine='heuristic'` default = 5 regex rules, or `'semgrep'`). Findings persist on `nodes.security_tags`.
`report`	Re-emit cached findings as JSON (`format='json'`) or SARIF v2.1.0 (`format='sarif'`).
`suppress`	Suppress a finding by `rule_id` (or `remove=true` to un-suppress).
`rule_list`	List available rules for an engine.

The semgrep engine requires the [security] extra and runs Semgrep's p/auto registry pack plus a 3-rule curated overlay.

`help` -- Full documentation

Returns complete documentation for each tool. Use when the compressed descriptions above are insufficient.

`config__open_relay` -- Re-trigger the relay setup form

Registered automatically from mcp-core. In HTTP mode it returns <PUBLIC_URL>/authorize so the agent can re-open the browser setup form (e.g. after credential expiry); in stdio mode it returns status: 'stdio_unsupported'.

CLI

The package installs two console scripts: crg (primary) and better-code-review-graph (legacy long name). Running either with no arguments starts the MCP server over stdio; a leading positional argument routes to a local CLI subcommand that calls the same domain services used by the MCP adapter. Run them directly after pip install, or without a persistent install via uvx --python 3.13 --from better-code-review-graph crg ....

# Start the MCP server over stdio (default -- no subcommand)
crg

# Build, inspect, and embed the local graph
crg graph build
crg graph stats
crg graph embed

# Query relationships and impact
crg query query \
  --pattern callers_of --target "path/to/module.py::function"
crg query search --search-query "authentication"
crg query impact --changed-files src/app.py

# Produce review context and run a local security scan
crg review context --base HEAD~1
crg security scan --engine heuristic

Command	Description
`graph build`	Full or incremental graph build. `--full-rebuild` re-parses every file; `--base <ref>` sets the incremental diff ref; `--repo-root <path>` overrides auto-detection.
`graph embed`	Compute vector embeddings using local ONNX or the configured cloud chain.
`graph stats` / `graph export` / `graph import` / `graph summarize`	Inspect, export/import a portable `crg` graph, or summarize functions.
`query query` / `query search`	Run relationship patterns or keyword/semantic search.
`query impact` / `query large_functions`	Analyze changed-file blast radius or find oversized nodes.
`query spot_check` / `query renamed_in_diff` / `query diff`	Inspect callsites, line shifts, or commit-to-commit graph changes.
`review context` / `review delta`	Generate review context or diff buckets for a code change.
`security scan` / `security report` / `security suppress` / `security rule_list`	Run and manage heuristic/Semgrep security findings.
`config status` / `config delete`	Show or remove stored credential config (`--yes` skips confirmation).
`doctor`	Environment self-check from shared `mcp-core` CLI.
`relay status` / `relay open` / `relay reset`	Inspect, open, or clear the relay setup session.

CLI subcommands print structured JSON and exit non-zero on an error. The config, doctor, and relay subcommands come from shared mcp-core.

Features

What this fork fixes versus the upstream code-review-graph:

Feature	code-review-graph	better-code-review-graph
Multi-word search	Broken (literal substring)	AND-logic word splitting
callers_of/callees_of	Empty results (bare name targets)	Qualified name resolution + bare fallback
Embedding	sentence-transformers + torch (1.1 GB)	fastretrieval ONNX + cloud (200 MB), dual-mode
Output size	Unbounded (500K+ chars)	Paginated (max_results, truncated flag)
Tool design	9 individual tools	7 grouped tools: graph + query + review + config + security + help + config__open_relay
Plugin hooks	Invalid PostEdit/PostGit	Valid PostToolUse

Comparison

How better-code-review-graph stacks up against direct competitors in each pillar:

Capability	better-code-review-graph	Greptile	Sourcegraph (Cody / MCP)	CodeGraph (colbymchenry)
Codebase knowledge graph	Yes (Tree-sitter, 14 langs, SQLite)	Yes (functions/classes/deps)	Yes (precise code indexing)	Yes (Tree-sitter, 20+ langs, SQLite)
Persistent incremental updates	Yes (git-diff + file-hash re-parse)	?	Yes (continuous indexing)	Yes (OS file-watcher debounced)
Qualified call resolution (callers/callees)	Yes (same-file bare-call resolution + fallback)	?	Yes (go-to-def / find-references)	Yes (callers / callees / impact)
Semantic search / embeddings	Yes (fastretrieval local registry + cloud Jina/Gemini/OpenAI/Cohere)	?	Yes (semantic + keyword + regex)	No (FTS5 full-text only)
Token-optimized review context	Yes (`review` tool, git-diff scoped)	Yes (PR review comments)	No (code-context assistant)	No (context layer, not review)
Security scanning	Yes (Semgrep `p/auto` + 3-rule overlay, SARIF)	?	?	No
Self-hostable	Yes (stdio default, machine-bound)	Yes (Docker / K8s / air-gapped)	Yes (self-hosted instance)	Yes (100% local, no API keys)
Free / open source	Yes (Apache-2.0)	No (proprietary SaaS; free OSS tier)	No (Enterprise license, source private)	Yes (MIT)

Sources: Greptile · Greptile pricing · Sourcegraph MCP · CodeGraph. Cells marked ? are capabilities the competitor does not publicly document, not confirmed absences.

Security

Explicit selection -- Cloud embedding errors are reported; the runtime does not silently switch models or fall back to local ONNX.
Error handling -- Tools return error strings with fix suggestions, never crash.
Read-only mount -- Docker mode mounts the repo as :ro (read-only).
SSRF-guarded endpoints -- Custom EMBEDDING_API_BASE / LLM_API_BASE URLs are validated before any outbound call.

To report a vulnerability, see SECURITY.md.

Build from source

git clone https://github.com/n24q02m/crg
cd better-code-review-graph
uv sync --group dev
uv run pytest
uv run better-code-review-graph

Requirements: Python 3.13, uv.

Trust model

This plugin implements TC-Local (machine-bound, single trust principal). See the mcp-core trust model for full classification.

Mode	Graph DB	Cloud credentials	Who can read your data?
stdio (default)	`<repo>/.better-code-review-graph/graph.db` (git-ignored)	`~/.better-code-review-graph-mcp/config.json` (AES-GCM, machine-bound key)	Only your OS user
HTTP self-host (multi-user)	Per-user `~/.crg/subs/<sub>/graph.db`	Per-user `~/.crg/subs/<sub>/config.json`	Only the authenticated user

Migration & changelog

Graph, security scan cache, and suppression state now use the package-owned .better-code-review-graph/ directory. Run graph(action="build", full_rebuild=true) once after upgrading, followed by graph(action="embed") if semantic search is needed. The ambiguous old .code-review-graph/ and .code-review-graph.db paths and their SQLite sidecars are left untouched: they may belong to the separate upstream package. Review and reapply any desired suppression rules explicitly.

The v2.0 release added temporal columns (valid_from_sha / valid_to_sha on every node and edge) plus an opt-in security scanner. The schema migration is auto-applied on first GraphStore open, and a backup of the pre-2.0 DB is written to <graph_db>.pre-2.0.bak. To downgrade and restore it:

CRG_DOWNGRADE_TO_1_X=1 uvx better-code-review-graph

Full schema-change list, behavior changes, and rollback procedure: BREAKING_CHANGES.md. Release-by-release history: CHANGELOG.md.

Documentation

Full docs at mcp.n24q02m.com/servers/better-code-review-graph/setup/:

Setup -- install methods for Claude Code, Codex, Gemini CLI, Cursor, Windsurf, mcp.json
Modes overview -- stdio / local-relay / remote-relay / remote-oauth
Multi-user setup -- per-JWT-sub credential model

Use the help tool from any MCP client for inline per-tool reference.

License

Apache-2.0 -- See LICENSE.

better-code-review-graph

Documentation

Better Code Review Graph

v2.0 migration (BREAKING)

Table of contents

Install

Local-first boundary

Smithery

Configuration

Model selection

Provider API keys

Advanced

Example -- cloud embeddings + summaries

Workspace username (HTTP setup form)

Tools

graph -- Graph lifecycle

query -- Graph queries

review -- Code review context

config -- Server configuration and credential setup

security -- Security scanning

help -- Full documentation

config__open_relay -- Re-trigger the relay setup form

CLI

Features

Comparison

Security

Build from source

Trust model

Migration & changelog

Documentation

License

`graph` -- Graph lifecycle

`query` -- Graph queries

`review` -- Code review context

`config` -- Server configuration and credential setup

`security` -- Security scanning

`help` -- Full documentation

`config__open_relay` -- Re-trigger the relay setup form