Enum Error 

pub enum Error {
    Auth(String),
    RateLimit {
        retry_after: Option<Duration>,
    },
    InvalidRequest(String),
    ContextTooLong {
        max_tokens: u32,
        requested_tokens: u32,
    },
    ModelUnavailable(String),
    ServerError {
        status: u16,
        body: String,
    },
    UpstreamServerFault(String),
    ContentFilter(String),
    Network(Box<dyn Error + Send + Sync>),
    StreamInterrupted(String),
    Cancelled,
    Adapter(Box<dyn Error + Send + Sync>),
    StreamIdle(Duration),
}

All errors that can be produced by a caliban provider.

Variants

Auth(String)

Authentication credentials were missing or rejected.

RateLimit

The provider rejected the request due to rate limiting.

Fields

retry_after: Option<Duration>

How long to wait before retrying, if known.

InvalidRequest(String)

The request was structurally invalid.

ContextTooLong

The request exceeds the model’s context window.

Fields

max_tokens: u32

The model’s maximum context size.

requested_tokens: u32

The number of tokens in the request.

ModelUnavailable(String)

The requested model is not available via this provider.

ServerError

The provider returned an HTTP error response.

Fields

status: u16

HTTP status code.

body: String

Response body text.

UpstreamServerFault(String)

The upstream model server reported an internal fault (process crash, OOM kill, segfault, etc.) — distinct from ServerError because the fault may arrive in-band (HTTP 200 + SSE error payload, as LM Studio does when the model crashes mid-stream). The fault is server-side, not request-side, so callers should surface it as such rather than as InvalidRequest.

ContentFilter(String)

The response was blocked by a content-safety filter.

Network(Box<dyn Error + Send + Sync>)

A transport-level network error occurred.

StreamInterrupted(String)

The HTTP response body was severed mid-stream. Distinct from Network because the request itself succeeded (the upstream accepted it and started replying) — what failed was reading the streaming body to completion. Typical triggers: TCP RST or FIN from upstream while the SSE stream was in flight, idle teardown by NAT/proxy, or a transient connection reset. The wrapped string is the underlying transport error chain, captured at the point the chunk read failed.

Cancelled

The operation was cancelled before completion.

Adapter(Box<dyn Error + Send + Sync>)

An adapter-specific error that does not fit other categories.

StreamIdle(Duration)

The streaming response went silent past the idle timeout.

Implementations

impl Error

pub fn network(e: impl Error + Send + Sync + 'static) -> Self

Wrap a network-layer error.

pub fn adapter(e: impl Error + Send + Sync + 'static) -> Self

Wrap an adapter-specific error.

pub fn stream_interrupted(inner: impl Display) -> Self

Wrap an upstream-severed-stream error. inner is rendered into the message so the user-visible line reads “stream interrupted mid-response: ”.

Trait Implementations

impl Debug for Error

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Display for Error

fn fmt(&self, __formatter: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Error for Error

fn source(&self) -> Option<&(dyn Error + 'static)>

Returns the lower-level source of this error, if any.

fn description(&self) -> &str

👎Deprecated since 1.42.0:

use the Display impl or to_string()

fn cause(&self) -> Option<&dyn Error>

👎Deprecated since 1.33.0:

replaced by Error::source, which can support downcasting

fn provide<'a>(&'a self, request: &mut Request<'a>)

🔬This is a nightly-only experimental API. (error_generic_member_access)

Provides type-based access to context intended for error reports.