Skip to content
CAI
Software that uses CAICheck a score

tirth8205/code-review-graph

54.5

Adequate · 15 November 2026

89.5k

lines of production code

Python

primary language

5

measurements over time

CAI band scale
CAI trend line
CAI lens gauges

What this system is

Features

Introduce D3-based code review graph visualization in VS Code webview

The VS Code extension now renders a force-directed graph of code entities (files, classes, functions, tests, types) using D3.js within the webview. Nodes are distinguished by specific shapes (circle, square, triangle, diamond, cross) and colors mapped to their kind, while edges represent relationships like calls, imports, inheritance, and tests, each with distinct colors and arrow markers. The visualization supports zooming, panning, and filtering by edge type, and communicates with the extension host via postMessage for state management and interaction.

code-review-graph-vscode/src/webview · high confidence

Introduce MCP tools package with 30 code-review graph tools

The \code\_review\_graph/tools\ package now exposes 30 MCP tools for graph analysis, search, and review workflows. This includes entry points like \build\_or\_update\_graph\ and \get\_minimal\_context\, analysis tools for architectural hotspots (\get\_hub\_nodes\, \get\_bridge\_nodes\) and structural weaknesses (\get\_knowledge\_gaps\), community and flow exploration (\list\_communities\, \list\_flows\), and refactoring capabilities (\refactor\_tool\, \apply\_refactor\_tool\). The package also provides semantic search (\semantic\_search\_nodes\), documentation retrieval (\get\_docs\_section\), and cross-repository search (\cross\_repo\_search\), all structured with bounded responses and provenance metadata. Also in this area: 16 commits (12 fixes) fixing code\_review\_graph/tools; 1 commit (1 fix) fixing code-review-graph, code\_review\_graph.

_code\_review\graph/tools · medium confidence · unverified

Introduce VS Code backend for CLI interaction, SQLite reading, and file watching

The VS Code extension now includes a dedicated backend layer in \code-review-graph-vscode/src/backend\ to manage the Python CLI, read the generated graph database, and monitor file changes. The new \cli.ts\ module provides a \CliWrapper\ that detects available Python installers (uv, pipx, pip3), installs the \code-review-graph\ package, and executes graph operations (build, update, watch, embed) with configurable timeouts and progress notifications. The \sqlite.ts\ module introduces a read-only \SqliteReader\ using \better-sqlite3\ to query graph nodes, edges, and statistics, including retry logic for database opening and graceful handling of ABI mismatches. The \watcher.ts\ module adds a \GraphWatcher\ that debounces file system events on the \graph.db\ file to trigger updates without overwhelming the event loop.

code-review-graph-vscode/src/backend · high confidence

Introduce VS Code extension for code graph visualization and analysis

Adds a new VS Code extension that integrates with the code-review-graph backend to provide interactive code analysis features. Users can build and update code graphs, visualize call graphs and blast radius impacts through tree views and webviews, and navigate directly to source code locations. The extension supports multi-root workspaces, automatically detecting graph databases in \.code-review-graph/graph.db\ or \.code-review-graph.db\, and includes onboarding walkthroughs and status bar indicators for a guided experience. Also in this area: 7 commits (5 fixes) fixing code-review-graph-vscode.

code-review-graph-vscode, code-review-graph-vscode/src · medium confidence · unverified

Introduce VS Code extension with interactive graph and file tree views

The new \code-review-graph-vscode\ extension adds a VS Code webview for an interactive, force-directed code graph visualization (powered by D3.js) and a hierarchical file tree view. The graph view supports exporting to SVG or PNG, highlights specific nodes, and respects a configurable maximum node count to manage performance. The file tree view organizes source files and their symbols (functions, classes, types, tests) with clickable navigation to the relevant file locations, and displays outgoing and incoming dependency edges (calls, imports, inheritance, etc.) for each symbol. A status bar item provides a quick summary of the graph database state, including node counts and staleness warnings if the data is older than one hour. A separate 'Blast Radius' tree view is also available to display changed and impacted nodes based on impact radius analysis.

code-review-graph-vscode/src/views · high confidence

Introduce code-review-graph VS Code extension with graph-aware navigation and SCM integration

This change adds the VS Code extension for the Code Review Graph, introducing several new capabilities in the \code-review-graph-vscode/src/features\ directory. Users can now run a cursor-aware blast radius command to visualize the impact of changes up to a configurable depth, and use navigation commands to find callers and tests for the symbol under the cursor, including matching by graph edges and naming conventions. A new live-search command provides quick navigation to functions, classes, files, and types via a debounced QuickPick. Additionally, an SCM integration command analyzes staged and unstaged git changes to compute their blast radius, while a file decoration provider adds visual badges in the Explorer and SCM views to indicate whether changed files are impacted, fully tested, or untested based on the underlying graph database.

code-review-graph-vscode/src/features · high confidence

Introduce code\_review\_graph v2.3.8 with context-savings estimation and legacy instruction support

The package is updated to version 2.3.8 and now exposes a new \context\_savings\ module that provides functions to estimate and format token savings for codebase exploration. To ensure smooth upgrades, a new \\_legacy\instructions\ module stores verbatim copies of instruction blocks from previous releases, allowing the system to correctly identify and upgrade managed blocks in user configuration files without guessing boundaries. The package also includes a vendored D3.js v7.9.0 asset for offline graph visualization and adds a \\\main\\_.py\ entry point to allow running the tool via \python -m code\_review\_graph\. Also in this area: 1 commit (1 fix) fixing code-review-graph, code\_review\_graph; 67 commits (58 fixes) fixing code-review-graph; Exclude TypeScript interfaces and type aliases from test gap analysis; Optimized C header evidence scanning in code review graph; Config-driven custom language support.

code-review-graph · low confidence · unverified

New diagnostic and auditing scripts for Python imports, PyPI connectivity, and PR comments

Added four new scripts to the \scripts/\ directory to improve developer experience and CI reliability. \audit\_python\_imports.py\ validates Python import edges in the code-review-graph database against CPython's \ast\ module, specifically checking relative imports, cross-module calls, and case-sensitive file existence to catch resolution bugs. \diagnose\_pypi\_connectivity.py\ checks TLS and HTTP connectivity to PyPI to help troubleshoot installation failures in environments with restricted network access. \render\_pr\_comment.py\ generates risk-scored GitHub PR comments from analysis JSON, handling path relativization for CI runners and enforcing artifact size limits. \smoke\_google\_embeddings.py\ provides a quick verification that the Google embedding provider initializes correctly with its optional dependencies.

scripts · high confidence

New evaluation benchmark suite for code review graph performance and accuracy

The evaluation framework now includes a comprehensive set of benchmarks to measure the graph's capabilities. An agent baseline benchmark compares realistic grep-and-read costs against graph query efficiency. Impact accuracy is now evaluated using an honest co-change ground truth (excluding the seed file) alongside the previous graph-derived mode, with failures correctly excluded from aggregates rather than inflating scores. Token efficiency metrics now properly exclude failed tool calls from averages. Additional benchmarks measure build performance (flow and community detection timing), flow completeness (entry point recall), multi-hop retrieval accuracy (semantic search plus graph traversal), and search quality (MRR ranking). Also in this area: 1 commit (1 fix) fixing code\_review\_graph/eval.

_code\_review\graph/eval/benchmarks · medium confidence · unverified

New evaluation framework for benchmarking code-review-graph

The \code\_review\_graph/eval\ package introduces a dedicated evaluation framework to measure and report on the performance of graph-based code reviews. It provides scoring metrics for token efficiency, Mean Reciprocal Rank (MRR), and precision/recall, alongside a runner that orchestrates benchmarks across real repositories. The framework includes specific benchmarks for five workflows (review, architecture, debug, onboard, pre-merge) to measure token consumption, as well as benchmarks for impact accuracy, flow completeness, search quality, and incremental fidelity. Results are generated as detailed markdown reports and CSV files, allowing users to track improvements in context reduction and retrieval accuracy.

_code\_review\graph/eval · high confidence

Removals

Removal of server codebase components

The server module has been removed, including the entry points (\_\init\\.py, \\main\\_.py, main.py), the core logic (graph.py, incremental.py, parser.py, tools.py), and all associated data models and storage engines. This eliminates the ability to build, query, or maintain the persistent code knowledge graph via the MCP server.

server · high confidence

Behavioural changes

2 commits (0 fixes) modifying .beads

A change to existing behaviour in .beads — 2 commits, 9 files: “bd init: initialize beads issue tracking”; “chore: keep the issue tracker's local files out of git”.

.beads · medium confidence · unverified

Updated benchmark results for impact accuracy, multi-hop retrieval, and token efficiency

The evaluation suite has been refreshed with new benchmark data for May 25 and August 2, 2026, covering impact accuracy, multi-hop retrieval, and token efficiency across repositories including code-review-graph, express, fastapi, flask, gin, and httpx. The impact accuracy metrics now include a 'co-change' ground truth mode (excluding the seed file) alongside the existing 'graph-derived' mode, providing a more granular view of prediction precision. Additionally, the multi-hop retrieval benchmarks have been expanded to include new natural language query tasks for several frameworks, and token efficiency ratios have been recalculated to reflect current model performance.

evaluate · high confidence

Updated diagrams to reflect v2.3.4+ context savings metrics and new platform support

The diagram assets and their generation scripts have been refreshed to accurately visualize the estimated context savings metric introduced in v2.3.4, specifically demonstrating the \detect-changes --brief\ and \update --brief\ CLI surfaces and the \--verify\ flag for cross-checking against tiktoken. The visual comparisons now highlight the token reduction achieved by querying the graph versus reading the entire codebase, and the documentation has been updated to include Nix support and the newly added Kiro and Qoder platform integrations.

diagrams · high confidence

Updated hook configuration to use new event types and integrate code-review-graph

The hooks configuration has been updated to replace the deprecated PostEdit and PostGit events with SessionStart and PostToolUse. A new SessionStart hook now runs at the beginning of every session to check for the existence of the code-review-graph knowledge graph and provide guidance to the user on using graph-based tools (such as semantic search and impact radius) instead of manual file scanning. Additionally, PostToolUse hooks now automatically build the graph when entering a new worktree and update it after code edits or bash commands, ensuring the graph stays in sync with the codebase. Also in this area: 2 commits (1 fix) modifying hooks.

hooks · medium confidence · unverified

Fixes

7 commits (4 fixes) fixing code\_review\_graph/assets/code-review-graph

A fix in code\_review\_graph/assets/code-review-graph — 7 commits (4 fixs), 3 files: “fix(codex): address skill ownership and fallback review findings”; “docs: bring every doc in line with 2.4.0”; “feat(codex): install native CRG skill with safe CLI fallback”; and 4 more (refs \#1066).

_code\_review\graph/assets/code-review-graph · medium confidence · unverified

7 commits (5 fixes) fixing code-review-graph-vscode

A fix in code-review-graph-vscode — 7 commits (5 fixs), 9 files: “fix(vscode): watch mode keeps running until you stop it”; “fix(vscode): walkthrough build reports failure, and builds have no time limit”; “docs: bring every doc in line with 2.4.0”; and 4 more. Also in this area: Added tests for the SqliteReader module.

code-review-graph-vscode/test · medium confidence · unverified

Test coverage

Expanded test fixtures for multi-language parser coverage; Test infrastructure for isolation, determinism, and corpus baselines.

Dependencies

VS Code extension and Python core dependency updates

The VS Code extension (code-review-graph-vscode) now uses a new package-lock.json with dependencies including d3 ^7.9.0 and better-sqlite3 ^12.4.1, targeting VS Code ^1.85.0. The Python core (pyproject.toml) updates fastmcp to \>=3.2.4,\<4 to address [CVE redacted]/62801/66416 and Windows stdio EOF bugs, bumps mcp to \>=1.0.0,\<3, and adds upper-bound constraints to tree-sitter, networkx, watchdog, and other dependencies to prevent breaking changes from next major releases.

(dependencies) · high confidence

Housekeeping

Code Review Graph v2.3.9 release

This update releases version 2.3.9 of the Code Review Graph MCP server. The package now includes a vendored D3.js v7.9.0 library for offline graph visualization support, and the module exposes context savings estimation functions (such as \estimate\_context\_savings\ and \format\_context\_savings\) in its public API. Also in this area: 42 commits (34 fixes) fixing code\_review\_graph; Code review graph now routes by question type rather than being limited to review tasks; 1 commit (1 fix) fixing code-review-graph, code\_review\_graph.

_code\_review\graph · medium confidence · unverified

Written by cca.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

How this codebase got here

Score

  • CAI 56 → 55 (-1.0)
  • Rubric changed (rubric-2026.11.1 → rubric-2026.11.4) — scores are not directly comparable.

Lenses

  • Code Health 43 → 40 (-3.1)
  • Architecture 91 → 91 (-0.1)
  • Maturity 75 → 79 (+4.0)
  • Readiness 57 → 55 (-1.2)
  • Security 66 → 69 (+3.0)
  • Performance 100 → 100 (+0.0)

Resolved (89)

  • Change coupling: communities.py ↔ flows.py (code_review_graph/communities.py)
  • Critical vulnerability: [GHSA redacted] (uv.lock)
  • Duplicated block (10 lines × 3) (code_review_graph/graph.py)
  • Duplicated block (10–11 lines × 3) (code_review_graph/tools/build.py)
  • Duplicated block (12–13 lines × 3) (code_review_graph/skills.py)
  • Duplicated block (13 lines × 2) (code_review_graph/graph.py)
  • Duplicated block (14–15 lines × 2) (code_review_graph/parser.py)
  • Duplicated block (15 lines × 2) (code_review_graph/cli.py)
  • Duplicated block (15–16 lines × 2) (code_review_graph/uninstall.py)
  • Duplicated block (16 lines × 2) (code_review_graph/tools/build.py)
  • Duplicated block (16–18 lines × 2) (code_review_graph/parser.py)
  • Duplicated block (17–18 lines × 2) (code_review_graph/graph.py)
  • Duplicated block (17–18 lines × 2) (code_review_graph/parser.py)
  • Duplicated block (17–19 lines × 5) (code_review_graph/parser.py)
  • Duplicated block (18 lines × 2) (code_review_graph/changes.py)
  • Duplicated block (20 lines × 2) (code_review_graph/cli.py)
  • Duplicated block (20 lines × 2) (code_review_graph/skills.py)
  • Duplicated block (20–21 lines × 2) (code_review_graph/skills.py)
  • Duplicated block (30–31 lines × 2) (code_review_graph/flows.py)
  • Duplicated block (42–43 lines × 2) (code_review_graph/spring_resolver.py)
  • …and 69 more

New (292)

  • CodeParser._add_html_script_import (cognitive 32) (code_review_graph/parser.py)
  • CodeParser._add_xslt_apply (cognitive 23) (code_review_graph/parser.py)
  • CodeParser._add_xslt_apply (cyclomatic 20) (code_review_graph/parser.py)
  • CodeParser._apply_python_receiver_evidence (cognitive 16) (code_review_graph/parser.py)
  • CodeParser._apply_typed_call_targets (cyclomatic 16) (code_review_graph/parser.py)
  • CodeParser._emit_laravel_endpoint_nodes (cognitive 48) (code_review_graph/parser.py)
  • CodeParser._emit_laravel_endpoint_nodes (cyclomatic 32) (code_review_graph/parser.py)
  • CodeParser._extract_hcl_constructs (cognitive 21) (code_review_graph/parser.py)
  • CodeParser._extract_hcl_constructs (cyclomatic 18) (code_review_graph/parser.py)
  • CodeParser._extract_html_scripts (cognitive 27) (code_review_graph/parser.py)
  • CodeParser._extract_html_scripts (cyclomatic 16) (code_review_graph/parser.py)
  • CodeParser._extract_imports (cognitive 19) (code_review_graph/parser.py)
  • CodeParser._extract_js_var_functions (cyclomatic 22) (code_review_graph/parser.py)
  • CodeParser._extract_plsql_calls (cognitive 19) (code_review_graph/parser.py)
  • CodeParser._get_java_method_and_receiver (cyclomatic 18) (code_review_graph/parser.py)
  • CodeParser._index_ambiguous_method_scopes (cognitive 17) (code_review_graph/parser.py)
  • CodeParser._js_wrapped_function (cognitive 22) (code_review_graph/parser.py)
  • CodeParser._js_wrapped_function (cyclomatic 20) (code_review_graph/parser.py)
  • CodeParser._liquid_statements (cognitive 16) (code_review_graph/parser.py)
  • CodeParser._parse_html (cognitive 47) (code_review_graph/parser.py)
  • …and 272 more

Changes since last survey

  • 300 commits — 116 feature/other, 184 fixes

By area

  • (repo) — 84 commits
  • (root) — 65 commits
  • code_review_graph/graph.py — 18 commits
  • code_review_graph/skills.py — 14 commits
  • code_review_graph/cli.py — 12 commits
  • code_review_graph/tools — 7 commits
  • code_review_graph/parser.py — 6 commits
  • code_review_graph/incremental.py — 5 commits
  • code_review_graph/visualization.py — 5 commits
  • .github/workflows — 4 commits
  • code-review-graph-vscode/src — 4 commits
  • code_review_graph/doctor.py — 4 commits
  • code_review_graph/communities.py — 3 commits
  • code_review_graph/csharp_resolver.py — 3 commits
  • code_review_graph/main.py — 3 commits
  • code_review_graph/npm_alias_resolver.py — 3 commits
  • code_review_graph/assets — 2 commits
  • code_review_graph/eval — 2 commits
  • code_review_graph/migrations.py — 2 commits
  • tests/test_http_multi_repo.py — 2 commits

Notable commits

  • fix: Merge branch 'fix/gstack-cli' into fix/gstack-review-all
  • fix: Merge branch 'fix/gstack-install' into fix/gstack-review-all
  • fix: Merge branch 'fix/gstack-review' into fix/gstack-review-all
  • fix: Merge branch 'fix/list-cli' into fix/gstack-review-all
  • fix: Merge branch 'fix/list-graph' into fix/gstack-review-all
  • fix: Merge branch 'fix/list-install' into fix/gstack-review-all
  • fix: Merge branch 'fix/list-tests' into fix/gstack-review-all
  • fix: Merge branch 'worktree-agent-ab16ca93456489abb' into fix/gstack-review-all
  • fix: Merge pull request #1173 from tirth8205/fix/173-install-preview
  • fix: Merge pull request #1205 from tirth8205/fix/gstack-review-all
  • fix: Merge pull request #1208 from tirth8205/fix/promotion-gate
  • fix: Merge remote-tracking branch 'origin/fix/1059-drop-unread-tables' into train/2026-11-09
  • fix: Merge remote-tracking branch 'origin/fix/1073-write-lock-all-writers' into train/2026-11-09
  • fix: Merge remote-tracking branch 'origin/fix/1117-cjs' into train/2026-11-09
  • fix: Merge remote-tracking branch 'origin/fix/1145-seeded-viz' into train/2026-11-09
  • fix: Merge remote-tracking branch 'origin/fix/311-sibling-repos' into train/2026-11-09
  • fix: Merge remote-tracking branch 'origin/fix/607-multi-repo-http' into train/2026-11-09
  • fix: Merge remote-tracking branch 'origin/fix/850-laravel-tests-for' into train/2026-11-09
  • fix: Merge remote-tracking branch 'origin/fix/925-stop-clears-health' into train/2026-11-09
  • fix: Merge remote-tracking branch 'origin/fix/979-elixir-clauses' into train/2026-11-09
  • …and 280 more

Architecture

  • Unchanged — 0 containers · 1 contexts · 0 edges

Written by cca.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

Survey your own repository

tirth8205/code-review-graph was measured the same way every project in this corpus was: the same rubric, at a pinned commit, with the result published in full. Point a surveyor at a repository you know and see whether you agree with it.

About this page

  • The score is its most recent published measurement, taken on 15 November 2026 at a pinned commit. It is not a live figure and does not change until the project is measured again.
  • Measured at commit bbdda4b9e4b4818a9f04d7ef96ac0e50da13f208 — the exact code this score is about.
  • Scored under rubric-2026.11.4 — the same rubric and the same method as every other entry in this index.
  • Measured by Canine Code Analyzer (cca.canine.dev) using cca measure-2026.11.73+preprod-52de90b74216.