A CLAUDE.md is just a markdown file at the root of your repo. Copy the content below into your own project's CLAUDE.md to give your agent the same context.
npx versuz@latest install seaql-sea-orm --kind=claude-mdcurl -o CLAUDE.md https://raw.githubusercontent.com/SeaQL/sea-orm/HEAD/CLAUDE.md# SeaORM Project Guidelines
This project uses **SeaORM 2.0**. AI models likely have SeaORM 1.0 in their training data -- some patterns have changed. Always follow the 2.0 patterns shown below.
## Quick Reference Links
- [Walk-through of SeaORM 2.0](https://www.sea-ql.org/blog/2025-12-05-sea-orm-2.0/)
- [Migration Guide (1.0 to 2.0)](https://www.sea-ql.org/blog/2026-01-12-sea-orm-2.0/)
- [New Entity Format](https://www.sea-ql.org/blog/2025-10-20-sea-orm-2.0/)
- [Strongly-Typed Column](https://www.sea-ql.org/blog/2025-11-11-sea-orm-2.0/)
- [Nested ActiveModel](https://www.sea-ql.org/blog/2025-11-25-sea-orm-2.0/)
- [Entity First Workflow](https://www.sea-ql.org/blog/2025-10-30-sea-orm-2.0/)
## Entity Definition (2.0 Format)
In SeaORM 2.0, entities use `#[sea_orm::model]` with relations defined directly on the `Model` struct. This replaces the 1.0 pattern of separate `Relation` enums and `Related` trait impls.
```rust
mod user {
use sea_orm::entity::prelude::*;
#[sea_orm::model]
#[derive(Clone, Debug, PartialEq, Eq, DeriveEntityModel)]
#[sea_orm(table_name = "user")]
pub struct Model {
#[sea_orm(primary_key)]
pub id: i32,
pub name: String,
#[sea_orm(unique)]
pub email: String,
#[sea_orm(has_one)]
pub profile: HasOne<super::profile::Entity>,
#[sea_orm(has_many)]
pub posts: HasMany<super::post::Entity>,
}
impl ActiveModelBehavior for ActiveModel {}
}
```
### Relation Attributes
```rust
// Has-One
#[sea_orm(has_one)]
pub profile: HasOne<super::profile::Entity>,
// Has-Many
#[sea_orm(has_many)]
pub posts: HasMany<super::post::Entity>,
// Belongs-To (explicit foreign key mapping)
#[sea_orm(belongs_to, from = "user_id", to = "id")]
pub user: HasOne<super::user::Entity>,
// Many-to-Many via junction table
#[sea_orm(has_many, via = "post_tag")]
pub tags: HasMany<super::tag::Entity>,
// Self-referential
#[sea_orm(self_ref, via = "user_follower", from = "User", to = "Follower")]
pub followers: HasMany<Entity>,
```
### Junction Table (Composite Primary Key)
```rust
#[sea_orm::model]
#[derive(Clone, Debug, PartialEq, DeriveEntityModel, Eq)]
#[sea_orm(table_name = "post_tag")]
pub struct Model {
#[sea_orm(primary_key, auto_increment = false)]
pub post_id: i32,
#[sea_orm(primary_key, auto_increment = false)]
pub tag_id: i32,
#[sea_orm(belongs_to, from = "post_id", to = "id")]
pub post: Option<super::post::Entity>,
#[sea_orm(belongs_to, from = "tag_id", to = "id")]
pub tag: Option<super::tag::Entity>,
}
```
## Strongly-Typed Columns (2.0)
Use `COLUMN` constant with typed fields instead of the untyped `Column` enum:
```rust
// 2.0 (preferred) -- compile-time type safety
user::Entity::find().filter(user::COLUMN.name.contains("Bob"))
// 1.0 (outdated) -- still works but prefer COLUMN
user::Entity::find().filter(user::Column::Name.contains("Bob"))
```
## ActiveModel Builder Pattern (2.0)
```rust
// Create with nested relations
let bob = user::ActiveModel::builder()
.set_name("Bob")
.set_email("bob@sea-ql.org")
.set_profile(profile::ActiveModel::builder().set_picture("Tennis"))
.insert(db)
.await?;
// Add has-many children
let mut bob = bob.into_active_model();
bob.posts.push(
post::ActiveModel::builder().set_title("My first post")
);
bob.save(db).await?;
// Many-to-many
let post = post::ActiveModel::builder()
.set_title("A sunny day")
.set_user_id(bob.id)
.add_tag(existing_tag)
.add_tag(tag::ActiveModel::builder().set_tag("outdoor"))
.save(db)
.await?;
```
## Entity Loader API (2.0)
```rust
// Load with relations in a single query
let bob = user::Entity::load()
.filter_by_email("bob@sea-ql.org")
.with(profile::Entity)
.with(post::Entity)
.one(db)
.await?
.expect("Not found");
// Nested relations (post -> comments)
let user = user::Entity::load()
.filter_by_id(12)
.with(profile::Entity)
.with((post::Entity, comment::Entity))
.one(db)
.await?;
```
## Schema Registry (Entity-First Workflow)
```rust
// Auto-create tables from entity definitions (dev/testing)
db.get_schema_registry("my_crate::*")
.sync(db)
.await?;
```
## Anti-Patterns -- DO NOT DO THESE
### 1. Do not specify `column_type` on custom wrapper types
When using `DeriveValueType` for custom types, the column type is inferred automatically from the inner type. Adding `column_type` is redundant and incorrect:
```rust
// WRONG -- do not annotate column_type on custom types
#[sea_orm(column_type = "Decimal(Some((10, 4)))")]
pub speed: Speed,
// CORRECT -- SeaORM infers the column type from the DeriveValueType inner type
pub speed: Speed,
#[derive(Clone, Debug, PartialEq, DeriveValueType)]
pub struct Speed(Decimal);
```
### 2. Use `Text` or explicit max length for long strings on MySQL/MSSQL
On MySQL and MSSQL, `String` maps to `VARCHAR(255)` by default. For strings that may exceed 255 characters, use `Text` or specify `StringLen::Max`:
```rust
// WRONG on MySQL/MSSQL -- silently truncates at 255 chars
pub description: String,
// CORRECT -- use column_type for longer strings
#[sea_orm(column_type = "Text")]
pub description: String,
// Also correct -- explicit max length
#[sea_orm(column_type = "String(StringLen::Max)")]
pub event_type: String,
```
Note: Postgre / SQLite uses unbounded string by default, so this is primarily a MySQL/MSSQL concern.
### 3. Missing `ExprTrait` import
Methods like `.eq()`, `.like()`, `.contains()` on `Expr` require the trait import in 2.0:
```rust
use sea_orm::ExprTrait; // required in 2.0
Expr::col((self.entity_name(), *self)).like(s)
```
### 4. Do not use removed or renamed APIs
| 1.0 (removed/renamed) | 2.0 (correct) |
|---|---|
| `.into_condition()` | `.into()` |
| `db.execute(Statement::from_sql_and_values(..))` | `db.execute_raw(Statement::from_sql_and_values(..))` |
| `db.query_all(backend.build(&query))` | `db.query_all(&query)` |
| `Alias::new("col")` for static strings | `Expr::col("col")` directly |
| `insert_many(..).on_empty_do_nothing()` | `insert_many([])` returns `None` safely |
### 5. Do not manually impl traits that `DeriveValueType` now generates
In 2.0, `DeriveValueType` auto-generates `NotU8`, `IntoActiveValue`, and `TryFromU64`. Remove manual implementations to avoid conflicts.
### 6. PostgreSQL: `serial` is no longer the default
Auto-increment columns now use `GENERATED BY DEFAULT AS IDENTITY`. If you need legacy `serial` behavior, use feature flag `option-postgres-use-serial` or `.custom("serial")`.
### 7. SQLite: integer type mapping changed
Both `Integer` and `BigInteger` map to `integer` in 2.0. The entity generator produces `i64` by default. Override with `sea-orm-cli --big-integer-type=i32` if needed.