litellm/litellm-rust/crates/secrets
yujonglee 7b60fd061a
refactor(rust): adopt one error discipline across litellm-rust with source chains and Report (#48042)
* build(rust): upgrade reqwest to 0.13 and move MCP out of litellm-http

The workspace moves to reqwest 0.13, so litellm-http no longer carries a
second reqwest just for rmcp. The mcp feature, its pool and its proxy
conversion go away, and the gateway hands rmcp a pooled NoRedirect client
that already carries the TLS, proxy and timeout settings

The Client test helpers collapse into for_test, a no-redirect client
without TLS or proxy settings, and wrap_for_test for the few tests that
need a hand-built reqwest client

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(http): honor transport settings for Rust WebSockets

The Responses WebSocket connected through tokio-tungstenite with its own
native-roots TLS, so it ignored the configured CA bundle, verify flag,
client certificate, proxies and timeouts. The handshake now goes through
a pooled reqwest client (ClientVariant::WebSocket, HTTP/1.1, no
redirects) and upgrades the response, so it shares the settings every
other outbound request uses

The upgrade response is checked for the Sec-WebSocket-Accept key, the
Upgrade and Connection headers and a subprotocol that was offered

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* refactor(http): build outbound requests as reqwest::Request

OutboundRequest was a hand-rolled copy of reqwest::Request. Routes now build
a reqwest::Request directly and RequestSigner reads the body bytes back out
of it, so the signer still covers exactly what is sent. SigV4 signing moves
onto HeaderMap and takes the method, which drops the BTreeMap plumbing and
the header casing table

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* refactor(http): make litellm_http::Error the one opaque error

transport::Error and media::Error fold into litellm_http::Error, an
Arc<Repr> newtype whose private variants keep the typed reqwest, io,
rustls and tungstenite causes and own every message. Callers can no
longer build an error with a message of their own. They ask through
dispatched, is_request, status, upstream_response and tls_source

One From<reqwest::Error> now classifies connect, network and timeout
and strips the URL. The WebSocket handshake reports each failed check
as its own variant and owns its connect timeout. OCR drops the media
variants it mirrored, and inference::Error maps a header the signer
computes to Unsupported for every route, not only chat

The error rules move out of AGENTS.md into the rust-error skill, which
also records the decisions behind this design

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* docs(rust): split rust-error skill into thin SKILL.md with references

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* refactor(rust-error): render causes once and ask predicates at the boundaries

The four WebSocket handshake variants collapse into Handshake(HandshakeFault), and media download failures no longer answer dispatched() since they happen before the provider call

ErrorDetail in llms and auth-types, the inference error and the OCR error now store their cause in a field named cause with no #[source], so to_string() carries it exactly once and source() is None. Tests that walked the chain now destructure the variant and downcast the cause

The gateway asks is_unsupported() and is_missing_credential() on the inference and OCR errors instead of matching through nested enums, and litellm_auth::Error answers is_missing_credential() for both

The rust-error skill gains axum and pyo3 references, boundaries.md describes the real predicate order and the known divergences, the testing example asserts dispatched() correctly, and shape, classification, conversions, messages-and-causes and recipes agree with the code again

* docs(rust): focus the rust-error gateway page on the error flow and add a rust-gateway skill

The former axum page toured gateway-inference: its status match, the envelope
table, an extractor, the unsupported route and the Sse closure. Those details
belong to the gateway crates and go stale with them. The page is now gateway.md
and follows an error from the provider failure through each From into
IntoResponse, where it is classified once and rendered by surface and framing.

The gateway-structure material moves to a new rust-gateway skill: envelopes per
API surface, extractors and path registration, SSE relay, and the router test
harness.

* test(inference): pin provider URL redaction on every route's transport error

Chat, Messages, Responses and transcription each get a route-level test that
fails if a reqwest error reaches the caller with its URL. The Messages stream
decoder also takes body errors through litellm_http::Error like every other
reqwest error

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* docs(rust-error): adopt std source chains and add litellm-error Report

Messages describe their own layer and the cause is the source, boundaries
render the chain once with litellm_error::Report. Corrects skill claims
the audit found stale against the code.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* refactor(rust-error): keep causes as sources across http, auth, secrets, cache and utility crates

Messages describe their own layer and every cause is a typed source,
predicates match exhaustively, and boundaries render with
litellm_error::Report. Media downloads no longer count as dispatched,
Error::upstream bounds the provider body itself, and a forwarded
header SigV4 cannot sign is the caller's fault.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* refactor(rust-error): keep causes as sources in llms and the gateway crates

ErrorDetail loses its string variant, OCR provider variants become data,
host faults stop counting as the caller's, a tunnelled transport error
classifies the same on every stream path, and each gateway crate has one
Error with one status() and extractors that reject in its envelope.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* refactor(rust-error): nest llms errors in inference and ask predicates in route tests

litellm_inference::Error nests litellm_llms::Error instead of mirroring
its variants, host driver faults get their own variant, every predicate
matches exhaustively, and route tests ask predicates instead of matching
transport variants.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* refactor(rust-error): map bridge and gateway-inference errors once and render them with Report

RustUpstreamError has one constructor that always sets headers, argument
checks raise typed errors mapped once, OCR asks upstream_response(), the
Responses stream error event comes from the error type, gateway-inference
loses its string variants, and auth faults classify through one
litellm_auth predicate on every route.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(rust-error): carry OCR provider headers as a HeaderMap and keep outages off the caller

The OCR provider error and get_error_class take reqwest's HeaderMap, the
bridge turns it into Python's (name, value) pairs once. Credential
acquisition and provider SDK failures are no longer the caller's fault,
the UI hides auth and backend causes, and chat keeps a provider answer
as is.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* docs(rust-error): correct skill claims against the aligned code

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* refactor(rust-error): fix the remaining sweep findings in gateway, llms, auth and cache crates

Auth rejections hide their backend cause on every inference renderer,
path parameters reject in the surface envelope, Textract's unsupported
document and Azure DI's unparseable poll URL get typed variants, the
credential chain exposes its last failure as the source, gateway-mcp
asks predicates over rmcp errors, and the S3 and Azure blob transports
classify through litellm_http::Error.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* refactor(rust-error): fix the remaining sweep findings on the Python side

host-python keeps a native error's chain and returns blocking-pool
cancellation instead of panicking, Vault transport failures convert
through litellm_http::Error so no URL reaches Python, the bridge asks
predicates instead of matching three levels into Vault errors, and the
last string-built PyErrs become typed errors mapped once.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* build(deps): refresh compatible dependencies flagged by OSV

* refactor(rust): remove redundant client construction comment

* test(rust): cover configured MCP proxy and CA bundle

* fix(http): preserve native WebSocket trust and optional protocols

* fix(http): allow plain WebSockets without native roots

* fix(http): normalize provider URLs before building outbound requests

* fix(rust): preserve OCR download status and reduce error size

* fix(rust): preserve upstream bodies and bound diagnostics only

* test(rust): preserve bridge serialization boundary in body regression

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-10-28 09:22:11 -07:00
..
src refactor(rust): adopt one error discipline across litellm-rust with source chains and Report (#48042) 2026-10-28 09:22:11 -07:00
tests refactor(rust): adopt one error discipline across litellm-rust with source chains and Report (#48042) 2026-10-28 09:22:11 -07:00
Cargo.toml refactor(rust): adopt one error discipline across litellm-rust with source chains and Report (#48042) 2026-10-28 09:22:11 -07:00
PARITY.md test: move offline logging, secret manager and provider tests from legacy dirs into tests/unit (#45554) 2026-10-09 09:48:10 -07:00
README.md test: move tests/test_litellm core utils, routing, responses, caching and rust_bridge into tests/unit (#43199) 2026-09-25 17:10:13 -07:00

Secret resolution

Construct SecretManagerState::new(backend, settings) for a configured manager or use SecretManagerState::default() for environment lookups. The configured backend determines its provider identity. Write-only settings and names excluded by hosted_keys use the environment directly. secret_manager_would_be_consulted follows the same routing decision as resolution

Native resolution distinguishes a found value, confirmed absence, and a failed read. Missing values use the caller's default, while provider errors propagate. Empty strings are found values

new_python_compatible uses Python's environment fallback and conversion rules. Manager exceptions fall back to the environment, including custom-manager exceptions. AWS missing secrets, failed HTTP reads, missing string payloads, and missing or empty primary secrets return None without fallback, matching Python. The standard Python HTTP handler wraps network timeouts in litellm.Timeout, which AWS reads also swallow. Invalid primary JSON still raises. An absent AWS primary JSON field and a successful Azure response without a value also remain None. Defaults do not replace these results. .with_failure_policy(FailurePolicy::Propagate) exposes manager failures explicitly instead. Cancellation and other Python BaseExceptions always propagate unchanged. Explicit OIDC references keep their own errors and never use these fallbacks

The getters follow Python's conversion policy. Environment values use case-insensitive, whitespace-trimmed boolean parsing. Manager strings become booleans only when Python literal evaluation yields a boolean; other strings retain their exact contents. Non-string manager results produce None. get_secret_str returns only strings, and get_secret_bool accepts booleans or strings containing true or false. A type mismatch returns None and does not activate fallback

Route integration

Inject Arc<dyn SecretSource> from litellm_secrets::source into route preparation. SecretResolver implements this interface and supports arbitrary names through its asynchronous get_secret_str. It applies the same manager selection, conversion, and fallback policy to every lookup

For synchronous provider transformations, call source.resolve(names).await during preparation and inject the returned Secrets snapshot. A snapshot contains only those names and never reads the process environment implicitly. Resolve runtime names through the source before invoking a synchronous transformation. OCR uses this pattern; other routes can adopt it as they are implemented

The Python bridge uses this shared source and resolver. The shared proxy initializer captures effective configuration, and directly constructed LiteLLM managers are adapted at the dispatch boundary, and the bridge retains a native backend per configured client. Python reads and Rust routes share that backend. Custom Python implementations remain external callbacks. Rollout policy controls whether the native binding is selected. Public provider reads and Vault/CyberArk mutations use this selection; AWS mutation bindings remain unfinished. Mutations update the same native cache used by public reads. Vault retains complete write response bodies and Python-compatible error dictionaries while verifying rotation with fresh reads. The Python boundary retains primary JSON until return conversion so Python JSON numbers, values, and exception details survive unchanged

Backend contracts

AWS Secrets Manager, Azure Key Vault, Google Secret Manager, Vault, and CyberArk implement BaseSecretManager for reads with an operation context. Foreign provider contexts are rejected before cache access or I/O. Writes and deletes use separate SecretWriter and SecretDeleter capabilities. CyberArk rotation writes and verifies the replacement while retaining the old alias because Conjur does not support deletion through this API

Shared rotation verifies that the replacement has the requested value before deleting the old secret. Same-name rotation keeps the replacement. AWS same-name rotation uses its version update API directly, matching Python

Backend reads preserve payload strings. Conversion belongs to the resolver. Google caches only successfully decoded payloads, so values agree before and after caching. Native reads do not cache absence or failures. Python-compatible Google reads preserve Python's negative cache and its always-read override. Resource-not-found responses indicate absence; authentication, permission, transport, and malformed successful responses remain errors

The HashiCorp Vault backend is enabled with the hashicorp feature and reads KV v2 values from HCP_VAULT_* environment variables. It supports static tokens, AppRole authentication, and TLS certificate authentication

Intentional differences from Python

Native backends consistently distinguish absence from failure instead of swallowing provider errors. Python-compatible resolution maps these results back to the Python handler contract before applying fallback

hosted_keys excludes a name for every backend. Python's handler recognizes Azure SecretClient and Google KeyManagementServiceClient instances before the local branch, allowing excluded names to reach those providers. Rust treats that as a routing bug. test_rust_hosted_keys_exclude_azure_sdk_clients_too in tests/unit/rust_bridge/ocr/test_secrets.py pins this behavior

Google rejects malformed base64 and mismatched CRC32C values instead of accepting corrupted payloads. Python currently ignores the checksum and uses permissive base64 decoding. Rust follows RFC 4648 and Google's integrity guidance; failed_or_missing_reads_are_not_cached covers rejection and recovery

CyberArk cached reads preserve the original secret text. Python's shared cache attempts JSON decoding, so a secret such as "password" changes to password after the first read, and true changes to a Boolean. This corrupts the stored credential representation. test_public_cyberark_reads_reuse_authentication_and_cached_values demonstrates the Python defect and verifies stable native results

Test parity

The Python test inventory maps each secret-manager test to Rust coverage or its owning boundary

AWS, Vault, and CyberArk keep client/authentication, reads, and writes/rotation in private provider modules. Their existing integration-test targets group configuration, read/cache, and write/rotation cases, with shared fixtures local to each target. Python-compatible dispatch and string coercion live separately from native dispatch