Growth.
$13 per seat / month
- Up to 1,000 components
- Weekly scan
- Single sign-on
- Package of 50 seats
Pricing
Your architecture
documents itself.
Code Swan reads your repositories and builds the catalog, the architecture graph, the API inventory and the resource map. Nobody writes them and nobody keeps them current. Your coding agents query all of it directly.
Scale.
$18 per seat / month
- Up to 3,000 components
- Continuous refresh
- SCIM and custom retention
- Package of 150 seats
Enterprise.
Custom pricing
- Custom component limits
- Self-hosted or your own model
- Data residency and SLA
- Unlimited seats
These prices are for our cloud (SaaS) deployment. Running Code Swan in your own infrastructure is priced separately, see deployment.
Deployment Run it in your own infrastructure. Not priced here. The shape of the deal depends on your environment. Learn about self-hosted →
Self-hosted Code Swan runs in your own cloud account. Your code never leaves your perimeter.
Bring your own model Scanning runs on your own LLM keys and your existing provider contract.
Talk to us Every plan
The same catalog, at every size.
No part of the catalog is held back for a larger plan. What changes further up is how much of your estate is covered, how often it refreshes, and how you manage access.
- Living software catalog Components, owners, lifecycle and criticality, read from your repositories.
- Architecture graph C1, C2 and C3 views built from the dependencies your code actually has.
- API intelligence Every REST, GraphQL and gRPC surface your services expose, with live OpenAPI.
- Cloud resource topology The databases, buckets, queues and topics your code touches.
- Feature documents How each capability works across services, step by step, as built.
- MCP server Claude, Cursor and your own agents query the catalog directly.
- Semantic search Find a service by what it does, not by its name.
- Owner detection Read from contribution history. Ownership is observed, not declared.
Fine print
The parts worth reading.
Seats
- Any authenticated user counts as a seat: dashboard, API, or an MCP connection.
- Service accounts count too. CI, bots, internal tooling.
- Seats come as a package. The plan costs the same whether you fill it or not.
Components
- Anything you deploy or publish on its own.
- APIs, endpoints, resources and features come with it, never counted.
- Most teams carry ten to twelve per engineer. Plans allow about twenty.
Terms
- Twelve month commitment, billed monthly.
- No metered overage. Past your limit you move up a plan.
- Departments: three on Growth, ten on Scale, unlimited on Enterprise.
Questions
The details.
Why do coding agents count as seats?
Because that is where the work happens. An engineer whose agent queries the catalog fifty times a day is getting the product, whether or not they ever open the dashboard.
Counting only logins would price the past.
How many components will I have?
More than the number of services you are picturing. Libraries, workers, frontends, functions and pipelines each count as one.
Most teams land around ten to twelve per engineer. We tell you the exact number from your first scan, before you pick a plan.
How long does setup take?
Connect your version control with read-only access and the first scan starts immediately. Most teams have a complete catalog within hours.
Nothing to install, no changes to your repositories, no instrumentation to add.
Where does my code go?
We read your repositories to build the catalog. We never store your source. Secrets are redacted before anything is analysed, and your data is isolated per organisation at the database level.
If that is not enough for your security team, a self-hosted deployment puts the whole thing inside your own cloud account.