## typesafe_sdk.RetryPolicy

`dataclass`

```
RetryPolicy(
    max_retries: int = 2,
    backoff_initial: float = 0.5,
    backoff_max: float = 5.0,
    backoff_jitter: float = 0.25,
    http_statuses: set[int] = (
        lambda: {408, 429, *range(500, 600)}
    )(),
    respect_retry_after: bool = True,
    api_connection_error: bool = True,
    api_timeout_error: bool = True,
    exceptions: set[
        type[BaseException]
    ] = set(),
    predicate: Callable[[BaseException], bool]
    | None = None,
    timeout: float | None = 30.0,
)
```

Configuration for SDK retry behavior. Examples:

```
from typesafe_sdk import RetryPolicy, TypeSafeClient

client = TypeSafeClient(
    retry=RetryPolicy(
        max_retries=3, timeout=10.0, http_statuses={429, 500, 502, 503, 504}
    )
)
```

### max_retries

`class-attribute``instance-attribute`

```
max_retries: int = 2
```

Maximum retries after the initial attempt; `0` disables retries.

### backoff_initial

`class-attribute``instance-attribute`

```
backoff_initial: float = 0.5
```

First backoff delay in seconds, doubled each attempt up to `backoff_max`; zero disables backoff.

### backoff_max

`class-attribute``instance-attribute`

```
backoff_max: float = 5.0
```

Maximum backoff delay in seconds; zero disables backoff.

### backoff_jitter

`class-attribute``instance-attribute`

```
backoff_jitter: float = 0.25
```

Fraction of each backoff delay randomly subtracted, between 0 and 1.

### http_statuses

`class-attribute``instance-attribute`

```
http_statuses: set[int] = field(
    default_factory=lambda: {
        408,
        429,
        *range(500, 600),
    }
)
```

HTTP status codes that are retried.

### respect_retry_after

`class-attribute``instance-attribute`

```
respect_retry_after: bool = True
```

Whether to honor `Retry-After` and `retry-after-ms` response headers.

### api_connection_error

`class-attribute``instance-attribute`

```
api_connection_error: bool = True
```

Whether to retry `TypeSafeAPIConnectionError`, raised when the request cannot reach or read from the server.

### api_timeout_error

`class-attribute``instance-attribute`

```
api_timeout_error: bool = True
```

Whether to retry `TypeSafeAPITimeoutError`, raised when the request exceeds its timeout.

### exceptions

`class-attribute``instance-attribute`

```
exceptions: set[type[BaseException]] = field(
    default_factory=set
)
```

Additional exception types that trigger a retry, on top of the built-in rules.

### predicate

`class-attribute``instance-attribute`

```
predicate: (
    Callable[[BaseException], bool] | None
) = None
```

An optional predicate called with the raised exception; returning `True` triggers a retry in addition to the other rules.

### timeout

`class-attribute``instance-attribute`

```
timeout: float | None = 30.0
```

Total retry budget in seconds per SDK call, including the initial attempt and delays; `None` disables the limit. Stops before a retry whose delay would reach or exceed the budget, re-raising the last error.