How is this different than a knowledge graph memory system?

Memory systems rely on agents subjectively selecting what to remember. A code graph is an external structural model of your system, like a blueprint or a map. It's incrementally updated as code changes, accurate because it's derived directly from your source, and useful across agents and sessions without asking any agent to decide what was worth remembering.

How does the graph stay up to date as my code changes?

The graph updates incrementally. Each file change triggers a partial update, only re-processing changed files and their affected relationships, not a full rebuild. Agents always query an accurate model of the current state of your code.

Supermodel: Code-Graph API

Q: What's the difference between Supermodel and static analysis tools?

Static analysis tools are built for specific checks at a point in time. Supermodel builds a persistent, queryable multi-layer graph combining call relationships, dependency edges, domain classifications, and parse structure. The graph is designed to be queried by agents at runtime, stays current as code changes, and works across languages and repositories.

API Primitives

Ground truth for your codebase.

Our API provides ground truth built from static analysis. Dependency, call, domain, and parse graphs combined into a single multi-layer IR. Works across TypeScript, Python, Go, Rust, and more. The graph updates as code changes. Agents query it to understand dependencies, trace calls, map business domains, and avoid mistakes.

Function index Every function, class, method with location and signature

Call graph Cross-file and cross-package call relationships with line numbers

Dependencies Internal imports and external packages mapped and queryable

Multi-language TypeScript, Python, Java, Go, Rust, Ruby, C++, and more. One API, any language.

Git aware Recent changes, diffs, commit history linked to code symbols

Always current Graph updates as code changes. Query fresh data every time

Supported languages

TypeScript JavaScript Python Go Rust Java Ruby C++ Kotlin Swift + more

Supermodel MCP Server

Star

The Supermodel MCP server brings the full code graph API into your AI-native editor. Query any symbol, trace any dependency, understand any codebase directly from your agent, without leaving your workflow.

Claude Code Cursor Codex

10+ languages TypeScript, JavaScript, Python, Java, Go, Rust, Ruby, C++, Kotlin, Swift, and more

Multi-layer graphs Call graphs, dependency maps, domain classification, parse trees

Agent-native MCP protocol for direct integration with Claude, Cursor, Windsurf, and more

Explore MCP

Test Coverage Map

Star

AI guesses at test coverage. Graphs know. Test Coverage Map uses Supermodel's call graph to map every test file to the exact functions it exercises, surfacing untested code paths and showing you what breaks if a test disappears.

Graph-based mapping Traces actual call paths from test files to source functions, not line counts

Untested path detection Surfaces functions with zero test coverage across your entire codebase

Runs in CI Reports coverage gaps directly in pull requests as a GitHub Action

Try Test Coverage Map

Explore Code Visually with an IDE Extension

The Supermodel extension renders your code graph inside your editor so you can navigate callers, callees, dependencies, and ownership in context. Perfect for onboarding, reviews, and maintaining a birds-eye view of your architecture.

Choose your IDE

Cursor Windsurf VS Code

Supermodel IDE extension — interactive code graph

Codegraph Use Cases

Build on the graph.

A single API to generate and maintain code graphs across languages, files, and repositories: Trace calls, dependencies, imports, domain classifications and more. Here's what teams like yours are building.

See API Docs Browse all repos

Impact analysis

See every file affected by a change before you deploy. Query the graph to understand blast radius across your codebase.

→ Changed: auth/login.ts
✓ api/routes.ts (imports login)
✓ middleware/auth.ts (calls validateUser)
✓ tests/auth.test.ts (tests login flow)

CI optimization

Run tests only for changed dependencies. Skip everything else. Query the graph to build a dependency-aware test runner.

→ Modified: utils/format.ts
✓ Run: tests/utils.test.ts
✓ Run: tests/api.test.ts (depends on format)
⊘ Skip: tests/auth.test.ts (no dependency)

Dead code detection

Find unused functions and imports. Query the graph for symbols with no incoming calls. Auto-generate removal PRs.

→ Scanning: src/**/*.ts
✓ Found: formatLegacy() (0 callers)
✓ Found: OLD_CONFIG (0 references)
✓ Created PR: Remove unused symbols

Domain classification

Auto-discover business domains from code structure, no annotations required. Understand which files belong to which domain across any repo.

→ Analyzing: src/**
✓ Domain: Auth (12 files)
✓ Domain: Payments (8 files)
✓ Domain: Notifications (5 files)

Circular dependencies

Detect import cycles before they cause build failures or runtime errors. Trace the exact path of any circular dependency across modules.

→ Scanning: dependency graph
⚠ Cycle: auth.ts → user.ts
└──► session.ts → auth.ts
✓ Path length: 3 hops

Codebase visualization

Generate dependency graphs and architecture diagrams. Export to PNG, SVG, or ASCII for documentation and onboarding.

→ Visualizing: auth flow
login.ts ──► oauth.ts
└──► token.ts ──► db.ts
✓ Exported: auth-flow.svg

Architecture docs

Auto-generate architecture documentation from the graph. Run as a GitHub Action to keep docs in sync with your codebase.

→ Analyzing: repo structure
✓ 6 domains identified
✓ 47 cross-domain dependencies
✓ Published: repos.supermodeltools.com

Public graph API

Upload any repo. Get back call graphs, dependency maps, and domain classifications. Build what you want on top.

curl -X POST api.supermodeltools.com/v1/graphs
→ { "call_graph": { nodes: 847, edges: 2031 },
"dependency_graph": { modules: 94 },
"domain_graph": { domains: 6 } }

Pricing

Early adopter pricing. Sign up now, save forever.

Pro Typically $29

$19_/mo

✓ 3M LOC per month
✓ 25MB per upload

Start free trial

Growth Typically $299

$199_/mo

✓ 30M LOC per month
✓ 50MB per upload

Choose plan

Custom

✓ Unlimited LOC
✓ Custom upload size
✓ Custom integration
✓ Dedicated Slack channel
✓ VPC & self-host options

Contact Sales

Start your free trial

14-day free trial. Explore the API and tools before choosing a plan. Overage: $0.10 per 1,000 LOC.

Start free trial

Frequently asked questions

Memory systems rely on agents subjectively selecting what to remember. That works the same way human memory does: you try to hold onto the things that seem important, but you can miss things, and the selection is lossy by design. A code graph is more like a blueprint or a map. It's an external structural model of your system that exists independently of any conversation. It's incrementally updated as code changes, it's accurate because it's derived directly from your source, and it's useful across agents and sessions without asking any agent to decide what was worth remembering.

No. We process your code to build the graph and then immediately remove it. Your source code is not retained after processing. What we keep is the structural graph: function signatures, call relationships, dependency edges, domain classifications, and metadata. The actual source is never persisted.

No. Because your source code is removed immediately after processing, it is never available to train on. We only retain the structural graph, not the code itself. Training on something we don't store is not possible.

TypeScript, JavaScript, Python, Java, Go, Rust, Ruby, C++, Kotlin, Swift, and more. 10+ languages, one API. If your stack spans multiple languages, the graph works across all of them.

The graph updates incrementally. Each file change triggers a partial update — only the changed files and their affected relationships are re-processed, not a full rebuild. This keeps the graph current without meaningful performance overhead, so agents are always querying an accurate model of the current state of your code.

Supermodel is an HTTP API, so any agent that can make API calls can query it. We also provide an MCP server for Claude Code and other MCP-compatible clients, which lets agents query the graph as a native tool without any custom integration code. Full documentation is at docs.supermodeltools.com.

Static analysis tools like ESLint, SonarQube, or tree-sitter are built for specific checks at a point in time. Supermodel builds a persistent, queryable multi-layer graph that includes call relationships, dependency edges, domain classifications, and parse structure combined into a single representation. The graph is designed to be queried by agents at runtime, stays current as code changes, and works across languages and repositories. It's infrastructure for agents, not a linter.

If your business needs special compliance send us a message at engineers@supermodeltools.com.

Supermodel vs. the alternatives

	Supermodel	GitLab gkg	Sourcegraph	Baseline LLM
Cross-file dependency resolution	10+ and growing	Limited languages	Deep, mature	Guesses from context window
Works with any repo	Yes	Requires GitLab	Self-hosted or Cloud	Yes
Start using immediately	API key in minutes	Need GitLab	Contact sales	Yes
Feed structured context to AI agents	Uncompact	No	No	N/A
Build your own tools/products on the data	Public graph API	No	Private API	No
Dead code / test coverage analysis	Built-in tools	No	No	Unreliable
Runs in CI	GitHub Actions	GitLab CI only	No	Yes
Open source tools	Build your own	No (EE license)	No (proprietary)	Varies

Persistent context
for coding agents.

Ground truth for your codebase.

Graph-powered tools for your software factory

Uncompact

Supermodel MCP Server

Deadcode Hunter

Test Coverage Map

Explore Code Visually with an IDE Extension

Build on the graph.

Pricing

Start your free trial

Frequently asked questions

Supermodel vs. the alternatives

Your API key in minutes.

Persistent contextfor coding agents.

Ground truth for your codebase.

Graph-powered tools for your software factory

Uncompact

Supermodel MCP Server

Deadcode Hunter

Test Coverage Map

Explore Code Visually with an IDE Extension

Build on the graph.

Pricing

Start your free trial

Frequently asked questions

Supermodel vs. the alternatives

Your API key in minutes.

Persistent context
for coding agents.