Code Review Graph
Install a persistent codebase graph so review and change tasks load only blast-radius files instead of re-reading the whole repo.
Overview
Code-review-graph is an agent skill most often used in Ship (also Build) that maps your codebase in a local graph so the agent reads only blast-radius files during review and change tasks.
Install
npx skills add https://github.com/aradotso/trending-skills --skill code-review-graphWhat is this skill?
- Tree-sitter structural map persisted in local SQLite, exposed via MCP to the agent
- Queries blast radius of changes instead of full-project context—documented 6.8× fewer tokens on reviews, up to 49× on la
- Claude Code plugin or pip install with code-review-graph install for MCP registration
- Triggers cover incremental graph update, change-since-commit review, and affected-files discovery
- Optional embeddings extra for semantic search over the graph
- 6.8× fewer tokens on code reviews (claimed)
- Up to 49× fewer tokens on daily coding tasks in large monorepos (claimed)
Adoption & trust: 977 installs on skills.sh; 31 GitHub stars; 2/3 security scanners passed (skills.sh audits).
What problem does it solve?
Every coding and review session re-ingests the whole repository, burning tokens and missing which files actually depend on your diff.
Who is it for?
Solo builders on medium-to-large repos who use Claude Code with MCP and want cheaper, sharper reviews and blast-radius checks.
Skip if: Tiny single-file projects, teams that cannot run Python 3.10+/uv locally, or workflows that never use MCP-aware agents.
When should I use this skill?
User asks to build or set up code-review-graph, review with graph, analyze blast radius, reduce review tokens, incremental graph update, find affected files, or install the MCP server.
What do I get? / Deliverables
You get MCP-backed graph queries and smaller, targeted file reads so reviews and change analysis stay focused on affected code.
- Local SQLite code graph
- Registered MCP server for Claude Code
- Blast-radius or affected-file sets for a given change
Recommended Skills
Journey fit
Spans multiple journey phases - primary shelf plus alternate fits below.
Canonical shelf is Ship/review because the primary payoff is token-efficient code review and change analysis; graph setup also helps during Build when mapping dependencies. Review subphase matches blast-radius reads, incremental graph updates, and post-commit change analysis—the triggers explicitly target review and affected-files discovery.
Where it fits
After scaffolding a monorepo, build the graph once so later feature work queries dependencies instead of full-tree greps.
Before opening a PR, review changes since last commit using graph-limited file reads.
Estimate blast radius of a refactor to see which modules import the symbol you are changing.
Incrementally update the graph after hotfixes so production follow-up edits stay scoped.
How it compares
Skill plus MCP graph layer—not a replacement for linters or human judgment on security and design.
Common Questions / FAQ
Who is code-review-graph for?
Indie developers and small teams shipping with Claude Code who need structural codebase maps for review and dependency-aware edits without full-repo context every time.
When should I use code-review-graph?
Use it in Ship when reviewing changes since last commit or analyzing blast radius; in Build when setting up the graph for a new project or finding files affected by a change; anytime you want fewer tokens on coding tasks in large repos.
Is code-review-graph safe to install?
It indexes local source into SQLite on your machine; check the Security Audits panel on this page and treat the graph as sensitive if your repo contains secrets.
SKILL.md
READMESKILL.md - Code Review Graph
# code-review-graph > Skill by [ara.so](https://ara.so) — Daily 2026 Skills collection. `code-review-graph` builds a persistent structural map of a codebase using Tree-sitter, stores it in a local SQLite graph, and exposes it to Claude via MCP. Instead of re-reading entire projects on every task, Claude queries the graph and reads only the files in the blast radius of a change — averaging 6.8× fewer tokens on code reviews and up to 49× on daily coding tasks in large monorepos. --- ## Installation ### Claude Code Plugin (recommended) ```bash claude plugin marketplace add tirth8205/code-review-graph claude plugin install code-review-graph@code-review-graph ``` Restart Claude Code after installation. ### pip ```bash pip install code-review-graph code-review-graph install # registers the MCP server with Claude Code ``` Requires Python 3.10+ and [uv](https://docs.astral.sh/uv/). ### Optional: semantic search support ```bash pip install code-review-graph[embeddings] ``` Enables vector embeddings via `sentence-transformers` for `semantic_search_nodes_tool`. --- ## Initial Setup After installation, open your project in Claude Code and run: ``` Build the code review graph for this project ``` Or use the slash command: ``` /code-review-graph:build-graph ``` The first build parses the full codebase (~10 seconds for 500 files). After that, the graph updates incrementally on every file save and git commit (under 2 seconds for a 2,900-file project). --- ## CLI Reference ```bash # Register MCP server with Claude Code code-review-graph install # Parse entire codebase into the graph (first run) code-review-graph build # Re-parse only changed files (subsequent runs) code-review-graph update # Show graph statistics: node count, edge count, language breakdown code-review-graph status # Auto-update the graph as you save files (continuous watch mode) code-review-graph watch # Generate an interactive D3.js HTML visualisation of the graph code-review-graph visualize # Start the MCP server manually (Claude Code does this automatically) code-review-graph serve ``` --- ## Slash Commands in Claude Code | Command | What it does | |---|---| | `/code-review-graph:build-graph` | Build or rebuild the code graph from scratch | | `/code-review-graph:review-delta` | Review changes since the last commit | | `/code-review-graph:review-pr` | Full PR review with blast-radius analysis | --- ## MCP Tools (used automatically by Claude) Once the graph is built, Claude calls these tools without manual prompting: | Tool | Purpose | |---|---| | `build_or_update_graph_tool` | Build or incrementally update the graph | | `get_impact_radius_tool` | Find all files/functions affected by a change | | `get_review_context_tool` | Return a token-optimised structural summary for review | | `query_graph_tool` | Query callers, callees, tests, imports, inheritance | | `semantic_search_nodes_tool` | Search code entities by name or meaning | | `embed_graph_tool` | Compute vector embeddings for semantic search | | `list_graph_stats_tool` | Graph size and health statistics | | `get_docs_section_tool` | Retrieve documentation sections | | `find_large_functions_tool` | Find functions/classes over a line-count threshold | --- ## Configuration: Ignoring Paths Create `.code-review-graphignore` in the repository root: ``` generated/** *.generated.ts vendor/** node_modules/** dist/** __pycache__/** *.pyc migrations/** ``` The graph will skip these paths