Intelligent codebase visualizer.
Turn any JavaScript, TypeScript, React, Next.js, or Node.js repo into a living, queryable map — with functional summaries, technical summaries, and security analysis on every node.
Click the image to watch the demo
- About
- Who is this for
- Quick Start
- The Problem DevLens Solves
- Benchmarks
- Ways to Use DevLens
- Configuration
- What DevLens Understands
- Repository Layout
- DevLens Cloud
DevLens is a codebase visualizer for JavaScript, TypeScript, React, Next.js, and Node.js projects. It builds a typed dependency graph of every component, hook, function, route, and store — with AI-powered summaries on each node — so you can explore, understand, and analyze your architecture in seconds instead of hours.
What makes it different? Instead of reading files one at a time, you (or your AI agent) query the pre-built graph. A node summary costs ~50 tokens vs ~2,000 tokens per file.
- Onboard new devs in hours, not weeks — explore the graph instead of spelunking files.
- Review PRs with full context — see exactly what depends on each change.
- Run impact analysis before refactoring — "what breaks if I change this?"
- Catch circular deps, god-files, and coupling hotspots automatically.
- Keep living documentation — summaries stay fresh as code changes.
- Get a bird's-eye view of your entire codebase in seconds.
- Spot architectural debt before it becomes a crisis.
- Understand what your team has been building — even across repos.
- See how real codebases are designed — layers, patterns, data flow.
- Understand why things are connected, not just what each file does.
- Learn architecture patterns from production-grade projects.
Your agent burns tokens re-reading files it's seen before. DevLens gives it a graph to query instead.
Coming soon — DevLens Cloud: Shareable graphs your whole team can access, cross-repo navigation, and giving graphical context to your AI agents for smarter code review and analysis — all without running anything locally.
1. Install
npm install -g @devlensio/cliNo Node.js? Use the standalone binary installer (no dependencies):
Linux / macOS:
curl -fsSL https://raw.githubusercontent.com/devlensio/devlensOSS/main/scripts/install.sh | shWindows (PowerShell):
irm https://raw.githubusercontent.com/devlensio/devlensOSS/main/scripts/install.ps1 | iex
2. Init
cd your-project
devlens initThis sets up your LLM provider for AI summaries. Don't need AI? Skip this — structural analysis works offline.
3. Analyze
devlens analyze . --summarizeThis builds a typed dependency graph of every component, hook, function, route, and store — with AI-powered summaries on each node.
4. Explore
devlens overview # big picture: framework, stats, central nodes
devlens find-nodes -t ROUTE # find every route in the app
devlens security # see security flags across the codebaseAI coding tools let you ship faster than ever — but that speed creates AI debt: code merged without understanding, agents re-discovering connections every session, new hires drowning in unfamiliar structure.
DevLens fixes this by pre-building a typed dependency graph of your entire codebase. Every node gets:
- Functional summary — what business purpose does this serve?
- Technical summary — how does it work?
- Security assessment — severity + explanation
Armed with this graph, you (or your AI agent) can understand the full architecture in ~50 tokens per node instead of ~2,000 per file.
Tested across real-world tasks — architecture understanding, feature implementation, and bug finding — comparing the same model (DeepSeek V4 Flash, GLM 5.2, Kimi K2.6, Qwen 3.6) with and without DevLens.
| Metric | Without DevLens | With DevLens | Improvement |
|---|---|---|---|
| Avg cost per query | $0.163 | $0.075 | 54% cheaper |
| Avg input tokens | 88,980 | 35,035 | 61% less |
| Avg output tokens | 9,549 | 3,233 | 66% less |
| Avg tool steps | 14.3 | 7.8 | 45% faster |
| Structured architecture output | 50% | 100% | 2× more reliable |
| Architectural debt discovered | 0% | 50% | Now discoverable |
| Avg cache hit rate | 75.2% | 83.7% | +8.5pp |
Notable: Even the strongest model (DeepSeek V4 Flash) was 81% cheaper ($0.0035 vs $0.0185) and used 83% fewer input tokens with DevLens.
5 prompts across implementation and debugging tasks — DevLens graph context only (no per-node summaries).
| Task | Input tokens saved | Cache improvement |
|---|---|---|
| Continue Watching feature | 24% less input (56.8k vs 74.9k) | +5.2pp cache |
| Rate Limiting feature | 22% less input (32.1k vs 41.1k) | +8.3pp cache |
| Error Handling audit | (comparable) | Comparable |
| Profile Bug trace | 32% less input (36.9k vs 54.3k) | Similar |
When asked to explain a codebase's architecture:
| Capability | Without DevLens | With DevLens |
|---|---|---|
| Produced structured output | 50% | 100% |
| Referenced specific graph metrics | 0% | 100% |
| Identified architectural debt | 0% | 50% |
| Named specific important files | 75% | 100% |
Pick the interface that fits your workflow.
For when you want to see your codebase laid out as an interactive graph.
Open the Web UI, paste your repo path, and explore a force-directed canvas — click any node to see its summaries, callers, callees, and security flags. Search, filter, diff commits across versions.
git clone https://github.com/devlensio/devlensOSS.git
cd devlensOSS && bun install && bun run devFor scripts, CI, and when you want answers fast without leaving the terminal.
npm install -g @devlensio/cli| Command | Description | Example |
|---|---|---|
devlens analyze . |
Analyze a repository into a DevLens graph | devlens analyze ./my-app --summarize |
devlens overview |
Big picture — framework, stats, central nodes | devlens overview → "Next.js 15, 342 nodes, 12 routes, top node: Layout (9.2)" |
devlens blast-radius <nodeId> |
What breaks if I change this? | devlens blast-radius "src/auth.ts::login" → "14 dependents affected" |
devlens cycles |
Find circular dependencies | devlens cycles → Lists every circular import group |
devlens find-nodes -t <type> |
Filter nodes by type (ROUTE, COMPONENT, etc.) | devlens find-nodes -t ROUTE → Lists all 12 routes |
devlens find-nodes --severity high |
Find high-severity security issues | devlens find-nodes --severity high → "2 findings in auth module" |
devlens diff <from> <to> |
Compare two analyzed commits | devlens diff abc123 def456 → "Added: AnalyticsTracker, Changed: CheckoutForm" |
devlens security |
List every security issue | devlens security --min-severity high --json |
devlens config |
View or update configuration | devlens config --provider openrouter --model deepseek-v4-flash |
Each command supports --json for piping into scripts and CI pipelines.
Full reference:
src/cli/README.md— every command with examples and options.
The most powerful way to use DevLens. Your AI agent normally reads files one at a time — the DevLens Skill teaches it to query the pre-built graph instead.
npx @devlensio/skill installThen reload your tool and use /devlens in Claude Code, Cursor, or Kilo:
| Command | Description | Example |
|---|---|---|
/devlens architecture |
Full system overview — stack, modules, routes, patterns | /devlens architecture → Returns structured report: framework, 12 routes, 3 modules, security posture |
/devlens security-analysis |
Prioritized security findings with exploit notes | /devlens security-analysis high → "SQL injection in loginUser (reach: 14 nodes)" |
/devlens impact <symbol> |
Blast radius — what breaks if you change this? | /devlens impact loginUser → "14 dependents across 3 modules" |
/devlens tech-debt |
Circular deps, coupling hotspots, god-files | /devlens tech-debt → "3 cycles, Navbar has 28 dependents" |
/devlens guard [target] |
Warn before editing high-risk code | /devlens guard → " |
/devlens onboard |
Generate ONBOARDING.md for new devs |
/devlens onboard → Writes full onboarding doc to repo root |
/devlens explain [path] |
Understand a module with callers/callees | /devlens explain src/api/ → Walks through all API handlers |
/devlens find <name> |
Locate any component, hook, or function | /devlens find Button → "3 matches across components and tests" |
/devlens changes [range] |
Explain what changed and its impact | /devlens changes yesterday → "3 files, 2 features, 1 bug fix" |
/devlens diagram [type] |
Mermaid diagrams of architecture or flows | /devlens diagram architecture → Generates layered module diagram |
Full reference:
packages/skill-installer/README.md— all subcommands with examples, install options, and supported tools.
Wire DevLens into any MCP client (Claude Desktop, IDE plugins, etc.). Bundled inside the CLI — exposes 14 tools over the Model Context Protocol.
devlens mcp # stdio mode
claude mcp add devlens -- devlens mcp # register in Claude Code
devlens mcp http -p 7000 # HTTP modeEach MCP tool is a query into the pre-built graph — your agent can:
list_analyzed_repos— see what repos are already analyzedget_repo_overview— framework, stats, route count at a glancefind_nodes— search by name, type, file, severity, or scoreget_blast_radius— check impact before refactoringget_security_issues— rank all security findingslist_cycles— find circular dependenciesanalyze_changes— diff two analyzed commits
Full reference:
src/mcp/README.md— tool reference, examples, registration, and configuration.
Config lives in ~/.devlens/config.json — set via devlens init or devlens config.
| Provider | Recommended model | Notes |
|---|---|---|
| Ollama (local) | qwen2.5-coder:7b |
Free, local, 8GB+ RAM |
| OpenAI | gpt-4o-mini |
Fast, cost-effective |
| Anthropic | claude-sonnet-5 |
Excellent code understanding |
| OpenRouter | deepseek-v4-flash or mimo-v2.5 |
Best cost/quality balance |
| Gemini | gemini-2.0-flash |
Fast, large context |
devlens config --provider openrouter --model deepseek-v4-flash --api-key <key>Node types: COMPONENT, HOOK, FUNCTION, STATE_STORE, UTILITY, FILE, ROUTE, TEST, STORY, THIRD_PARTY
Route types: Next.js (app & pages), Express / Fastify / Koa, React Router / TanStack Router / wouter
Edge types: CALLS, IMPORTS, READS_FROM, WRITES_TO, PROP_PASS, EMITS, LISTENS, WRAPPED_BY, GUARDS, HANDLES, TESTS, USES, NEXTJS_API_CALL, NAVIGATES_TO
Per node: Importance score + functional summary + technical summary + security assessment
devlensOSS/
├── src/
│ ├── cli/ # `devlens` CLI (commander program + commands)
│ ├── core/ # Shared query core (CLI + MCP — never drift)
│ ├── mcp/ # MCP server (stdio + HTTP), 14 tools
│ └── server/ # Backend API for the Web UI
├── frontend/ # Next.js 15 graph visualizer (Cytoscape)
├── plugins/devlens/ # Agent Skill (Claude plugin source)
├── packages/skill-installer/ # @devlensio/skill — npx installer
├── bin/ # npm launcher
├── npm/<platform>/ # 5 prebuilt binary packages
├── scripts/ # Release tooling
└── server.json # MCP registry manifest
The analysis engine lives in the separate devlensio package.
A hosted version is in development:
- Shareable graphs your whole team can access
- Cross-repo navigation — understand your entire org
- Give graphical context to AI agents for smarter code review and analysis
- No local setup required






