## Why
Reusable clients need fresh authentication for each request while preserving the credential source and transport policy selected when the client was created.
## What changed
- Add runtime-aware backend client construction and per-request authentication. Allow token refresh within the original credential source, reject credential replacement, and validate destination and account/organization selectors.
- Let runtime authentication finish persisting rotated tokens when the requesting task is cancelled.
- Apply asynchronous authentication to file creation and finalization requests, generating fresh assertions while keeping credentials off blob uploads.
- Preserve direct sandbox routing without enabling legacy custom-CA fallback for protected clients that reject redirects.
- Exclude login account metadata from model cache identities when provider bearer credentials override the login, and record authentication selection outcomes and duration.
## Testing
Add regression coverage for credential refresh and replacement, organization selection, destination restrictions, transport policy preservation, cancellation during token rotation, credential stripping on redirects, and file upload authentication in both runtime authentication modes.
GitOrigin-RevId: d045595e869dfcdfd9f705a6228cf0eb4c2d2c12
## Why
Uploaded audio was rejected even for audio-capable models. Invalid or expired recordings also need to be excluded from later requests so users can continue the conversation.
## What changed
- Accept uploaded audio when the selected model supports audio; otherwise, preserve valid sibling input and report the unsupported attachment before inference.
- Track successful audio decoding by model and provider, and attribute provider rejections against the exact submitted request. Preserve attachments when attribution is ambiguous.
- Persist confirmed rejected file IDs across compaction and resume, omitting them from future requests without rewriting retained user history.
- Check audio availability before remote compaction and retry once without an unambiguously rejected recording. Include guidance to remove or replace affected attachments in audio errors.
## Testing
Add unit and scenario coverage for uploaded audio across follow-up turns, compaction, cold and paginated resume, model switching, ambiguous rejections, expired files, and queued or interrupted input.
GitOrigin-RevId: e701901e7b1c0865c9f2a5918197215f549b06c0
## What changed
Map provider `budget_reached` errors to `SessionBudgetExceeded` for HTTP 400 responses and streaming failures. End SSE streams immediately on this error, ignoring retry headers and later completion events, and display `budget reached` in the TUI.
## Testing
Add coverage for HTTP error classification, immediate SSE termination without retries, WebSocket error delivery and subsequent request recovery, and TUI error rendering and task-state cleanup.
GitOrigin-RevId: 6a62d699cdb9cb446214cacfacd0c9ed68bf21a6
## What changed
- Gate GIF output behind `features.image_generation_animation`, disabled by default.
- Accept `gif` options for generation and editing, with optional `frame_count` and `frame_duration_ms` controls. Use `gif: {}` for API defaults.
- Preserve animated output with `.gif` filenames and `image/gif` data URLs while providing a still preview to the model.
- Add guidance for simple animations and spritesheet-based GIF edits.
## Testing
Add integration coverage for GIF generation and editing, explicit frame controls and API defaults, saved animation bytes, and still-preview replay across follow-up turns.
GitOrigin-RevId: 08afa0316ae3675921652157836191a649f563cf
## What changed
- Add optional `ModelInfo.web_search_provider` metadata with `standalone` and `search_mcp` values.
- Add `tools.client_search_provider` and expose its value as `client_search_provider` in app-server configuration responses, with updated JSON schemas, TypeScript types, and Python bindings.
- Register the `hosted_search_policy` feature flag as under development and disabled by default.
GitOrigin-RevId: 263c3f81a61b50f8bb53ff338b9de8186db316e0
## What changed
Once Responses requests fall back to HTTP after exhausting WebSocket retries or receiving HTTP 426, keep subsequent requests and turns on HTTP for the remainder of the session. Clear cached WebSocket state and remove background recovery probes, recovery trials, and their acknowledgement timeout.
## Testing
Update fallback tests to verify that a second turn stays on HTTP without additional WebSocket attempts. Adjust retry and `Retry-After` request-count assertions to exclude background probes.
GitOrigin-RevId: a4f6376604527c14db8d60f79bd21c3aaf289f70
## What changed
Add `RealtimeSidebandTransport` and `Op::RealtimeConversationAttachStream` so callers can attach Core to an existing realtime call through bounded JSON channels without opening or reconnecting a socket.
Read events independently of outbound backpressure, enforce a 1 MiB message limit, and drain accepted input on detach or channel closure before flushing the transcript. Pending writes are never replayed.
## Testing
Add transport and Core integration tests covering delegation without a callback socket, stalled output, attachment replacement and detach, channel closure, oversized messages, and preservation of queued transcripts in history.
GitOrigin-RevId: 06b7d1dfe2388054438d72d925e510a959dfe106
## Why
A transient WebSocket failure previously kept the session on HTTP for its remaining lifetime, even if WebSocket connectivity recovered.
## What changed
- Probe WebSocket connectivity in the background during HTTP fallback, allowing a subsequent request to use a ready connection without delaying HTTP requests.
- Verify recovery with a real request and bound its first-response wait to five seconds. Keep HTTP 426 responses and recovery timeouts on HTTP for the rest of the session.
- Keep retries of an HTTP request on HTTP. If a recovered connection fails before output, replay the request over HTTP without consuming its ordinary retry budget; failures after output use normal retry accounting.
- Rate-limit recovery probes and honor `Retry-After` deadlines.
## Testing
Extend integration coverage for successful recovery within a turn, disconnects before and after output, full-context replay followed by incremental WebSocket requests, persistent HTTP 426 fallback, and retry deadlines.
GitOrigin-RevId: 1c139a4bbce1a67442db5622fd70ff0cc8056c9c
## Why
When a response fails with incomplete output, attempted tool calls can appear only in the terminal response payload. Failed inference traces previously omitted those items, losing evidence of the attempted call.
## What changed
Retain output items marked `incomplete` on `ResponseIncompleteError` and include recognized items in failed inference traces, deduplicating by item ID. Keep these items out of streamed output, tool dispatch, and the continuation cache, and omit their payloads from the error's debug representation.
## Testing
Add tests for retaining only incomplete output from terminal responses and recording attempted calls in failed traces without streaming them or creating tool calls. Cover unknown response item types being skipped.
GitOrigin-RevId: 594f6a37a5af543694f19fcffe0c164cb2addb36
## What changed
Map audio size, data, format, and sample-rate validation failures to `InvalidAudioInput` for HTTP 400 responses and streamed errors. Preserve the provider's message, falling back to `Invalid audio input.` when it is absent or blank.
Classify these errors as non-retryable and expose them as `BadRequest`. Redact their messages in doctor diagnostics and group them as invalid requests in telemetry.
## Testing
Add coverage for all six audio validation codes across HTTP errors, streamed `error` events, and `response.failed` events, checking error classification, message preservation, and the absence of retries. Extend doctor diagnostic redaction coverage for `InvalidAudioInput`.
GitOrigin-RevId: 0a59964bcabd83997bdb32fc9318a5482dac33b0
## What changed
Return `CatalogModel` from model lookups, keeping metadata together with its `Bundled`, `Provider`, `ExplicitConfig`, or `Synthetic` origin through config overrides and runtime snapshots. Replace `used_fallback_model_metadata` checks with explicit `Synthetic` origin checks.
Share the underlying `Arc<ModelInfo>` with extension stores and use copy-on-write for metadata changes. Keep origin tracking outside serialized model metadata.
## Testing
Extend coverage for origin retention across cache reloads, namespaced model lookups, catalog fallback, and config overrides. Verify that step snapshots and extension stores share the same model metadata allocation.
GitOrigin-RevId: bb6130c94a7a795aa168ee24664a1dd453579fba
## Why
Long realtime conversations could lose speech when active transcripts and handoff transcript fields were truncated.
## What changed
- Record completed V3 transcript chunks in model and client history once the active transcript exceeds 4 KiB, without triggering inference.
- Keep unfinished utterances pending to preserve conversation order, and persist completed chunks before subsequent handoffs.
- Preserve full V3 transcript text at handoff and session-end tail flush, and avoid adding handoff input already recorded in transcript history.
## Testing
Add integration coverage for long transcripts, XML escaping, recording before handoff without inference, repeated handoff input, and both session-end tail flush settings.
GitOrigin-RevId: 62218b811809c4cf5b16fb0247a335eb474f9ab9
## What changed
- Match final serving assessments to their response and output item, then retain them with tool invocations and code cells across waits.
- Use matching assessments instead of asynchronous scoring for adaptive review: low-risk actions can skip synchronous review, while elevated-risk actions require it.
- Revalidate captured authorization and policy before approval. Require fresh review for stale or incomplete authorization, preserve explicit review requirements, and discard assessments when hooks rewrite tool input.
- Keep asynchronous scoring as the fallback when an assessment is missing or does not match.
## Testing
Add integration coverage for direct and nested tool calls, hook rewrites, explicit escalation, interrupted responses, catalog-driven policy, incomplete authorization, and assessment retention across code-cell yields and waits.
GitOrigin-RevId: 4346a227204137f018c03c34b302d6f1e9592c96
## What changed
- Add the disabled-by-default `interrupt_on_cancel` feature to interrupt upstream inference when a Responses Lite turn is canceled, with a one-second cleanup deadline.
- Detach canceled WebSocket responses so replacement requests can open a new connection without waiting for cleanup.
- Add an optional `interrupt_on_cancel` override for thread samples, inheriting the parent setting when omitted.
## Testing
Add coverage for cancellation during inference and steering, replacement turns proceeding before cleanup finishes, and prompt cancellation of model response interceptors.
GitOrigin-RevId: 501368746fc04fd3fdbd45c1f34c75ce57f4e955
## Why
Stopping a sideband observer should close its WebSocket while preserving the frontend-owned realtime session. The existing frameless session close also sends `session.close`, so observer shutdown needs a transport-only close.
## What changed
- Add `close_transport()` to send a normal WebSocket close without `session.close`.
- Close the sideband transport on cancellation or when the call becomes inactive, including an existing connection awaiting adoption by the input task. Bound the close attempt to five seconds.
- Thread an optional sideband base URL through realtime start processing and allow nested thread ID fields in request serialization scopes.
## Testing
Add WebSocket and core conversation regression tests covering observer detachment for both realtime parsers and conversation versions. Verify that explicit frameless session close still sends `session.close`.
GitOrigin-RevId: 874a8980b0fcd45058e976b00c0d968f5b4b2404
## What changed
Decode `response.metadata` Guardian assessments into typed `GuardianAssessment` events over SSE and WebSocket, carrying `response_id`, `item_id`, and `requires_review`. Ignore assessments with missing, empty, or incorrectly typed required fields while preserving independent delivery of other metadata.
Expose the assessment type to request-scoped extension interceptors. Keep assessments out of conversation history; they do not authorize tool execution.
## Testing
Add SSE coverage for assessment ordering, malformed metadata, and mixed metadata delivery. Extend model request tests to cover SSE and WebSocket interceptor delivery and exclusion from subsequent model input.
GitOrigin-RevId: 282b8817ee97346d76b40bcb412eef023df60ff1
## Why
Shared routing cookies can change after a parent connection is established. Guardian reviews need the routing affinity captured with their parent response so unrelated traffic does not replace that hint.
## What changed
- Capture `__oailb` and `__cflb` cookies from successful ChatGPT WebSocket handshakes and pair them with the parent response ID.
- Retain the snapshot throughout a review and its retries, and apply it only to Guardian requests with the matching parent response. Reconnect cached reviewer sockets when the parent affinity changes.
- Preserve cookie expiry and restrict reuse to the original secure origin and endpoint path. Replace routing cookies while retaining unrelated cookies, and keep the captured affinity out of prompts, history, and serialized metadata.
## Testing
Add coverage for cookie scope, expiry, duplicate rejection, and merging; replaying a captured pin after shared cookies change or are deleted; and retaining the matching parent snapshot through compaction without exposing cookies in review prompts or metadata.
GitOrigin-RevId: d180679e6265b612a3617f20ca1f5dec89e73aa8
## Why
Handshake probes can lose upgrade evidence when a subsequent read fails, and diagnostic output can expose URLs, server error messages, and close reasons containing sensitive values.
## What changed
- Retain successful upgrade status and allowlisted tracking headers, including on post-upgrade read failures.
- Return structured probe errors that distinguish typed TLS failures from failures with an unknown stage while preserving the underlying API error.
- Recognize empty close frames as immediate closes with code `1005`.
- Redact wire values from probe debug output and doctor error details, and remove URLs and raw errors from connection logs.
## Testing
Add local TLS and WebSocket fixtures covering certificate rejection, ordinary I/O failures, successful and rejected upgrades, immediate closes, malformed frames, header retention, and redaction. Add doctor output snapshots and error-detail redaction assertions.
GitOrigin-RevId: 1ee53720a7a752043943170d8e4da205ec4e55f3
## What changed
Expose `literature_search_query` in standalone `web.run` when the active model lists it in `experimental_supported_tools`. Support query text, domain filters, and recency filters, and reject literature search calls for models without the capability.
Pass the active model information to extensions for each step and report literature queries as search actions in web search events.
## Testing
Add coverage for capability rejection, model-dependent schema exposure, app-server round trips with a turn-level model override, and a scenario that searches for a paper and opens the returned reference.
GitOrigin-RevId: 87ef1fadd80ca7bfc2a7642c4e443f9b9f2e0e00
## What changed
- Recognize `policy_capacity_unavailable` in HTTP and streaming API errors and expose it as `policyCapacityUnavailable` through the app-server protocol and generated bindings.
- Display “Policy-restricted capacity is unavailable. Please try again later.” while preserving overload retry behavior based on `Retry-After`.
- Handle the new error in TUI overload rendering, reviewer retries, and error telemetry.
## Testing
Add coverage for HTTP and streaming error classification, retry advice, WebSocket terminal error attribution, protocol serialization, and TUI rendering with and without an upstream message.
GitOrigin-RevId: 1b9536fba1e11cc891dcf689aee35410d82f5412
## What changed
Remove `AuthProvider::resolve_auth_headers` and use `apply_auth_headers` as the async interface for header-only authentication. Migrate providers, callers, and existing tests to the shared interface, passing an empty `HeaderMap` when only authentication headers are needed.
Document that providers may refresh credentials asynchronously and that callers must use the returned header map so providers can remove superseded credentials.
GitOrigin-RevId: 42449de6fddd2be2c50c475c8e92628493197007
## What changed
- Add opt-in runtime authentication through `ModelClient::with_runtime_auth`, requiring the provider and runtime manager to share the selected OpenAI credentials.
- Keep runtime identity, credentials, routing, and transport selection consistent for each request and its retries, signing a fresh assertion for each attempt.
- Apply authentication asynchronously to WebSocket handshakes, replace stale authentication headers, and reconnect when the runtime changes.
- Reject account, organization, or project selectors that conflict with the registered runtime, and preserve gateway OAuth composition.
## Testing
Add coverage for dual-header and assertion-only authentication, HTTP retries, runtime replacement and account switches, stale WebSocket headers, selector mismatches, gateway OAuth failures, transport revocation, and Realtime compatibility.
GitOrigin-RevId: 45b9efc2c80481406a0f6f6396a8cdd21d933c12
## What changed
Fetch catalogs configured with `model_catalog_url` without a response-body size limit, allowing responses larger than 1 MiB. Remove the oversized-response rejection test and update the catalog client documentation to drop the limit requirement.
GitOrigin-RevId: acd0dbbac5d548596b50faa309496de2db3588df
## Why
Provider errors with code `model_not_found` should end the turn immediately and preserve the provider's message instead of triggering retries.
## What changed
- Recognize `model_not_found` in HTTP 400, 403, and 404 responses and in streamed `error` and `response.failed` events.
- Preserve the provider message and any available HTTP status, and expose `modelNotFound` through the app-server protocol and generated schemas and SDK types.
## Testing
Add coverage for HTTP classification, statusless stream errors, and terminal failures without retries across HTTP, SSE, and WebSocket responses. Verify that `thread/read` preserves the error reported by the live notification.
GitOrigin-RevId: 88f824ac03c8c9669915b1e40c67b76c08326fbe
## Why
Remote registry requests need to recover when credentials are rejected or rotated while `exec-server` is running. Concurrent requests should reuse refreshed credentials rather than refresh the replacement token again.
## What changed
- Add an `AuthProvider` recovery hook and retry registry registration, connection, and harness-key validation once after a `401` when the provider supplies replacement credentials.
- Use live ChatGPT authentication for remote `exec-server` and guard proactive and unauthorized refreshes so concurrent callers share the completed refresh.
- Allow executor credentials to retry when the parent has already supplied a valid replacement.
- Keep connection authentication, refresh, and retry within the original connection deadline.
## Testing
Add coverage for concurrent refreshes, stored and parent-supplied credential rotation, refresh failures, retry limits, and connection deadlines, plus a CLI integration test for refresh and credential reload in a running executor.
GitOrigin-RevId: 9ed533e410762c7ac8b7c0c30bb93fd20d96778b
## Why
Synchronous auth header construction could silently return no credentials when managed authentication became unavailable. Callers need an error to avoid sending requests with missing or partial authentication.
## What changed
- Make `AuthProvider::add_auth_headers`, `to_auth_headers`, and auth header telemetry fallible, and propagate failures through request and WebSocket setup.
- Return a transient error when managed authentication is unavailable, and preserve network policy denials through auth error conversion.
- Retry transient MCP authentication errors while treating build errors and policy denials as terminal.
- Cancel local MCP event streams before resolving authentication so cleanup still works when credentials become unavailable.
## Testing
Add regression coverage for rejected partial authentication, async header resolution, terminal auth errors without sending requests, managed account changes, MCP retry classification, and local stream cancellation after authentication becomes unavailable.
GitOrigin-RevId: 39d981e232cba1e302de502a2fa6239101114ed1
## Why
Trimming interleaved realtime transcripts by utterance order can discard the latest speech from an earlier utterance. Evicting after each delta can also drop one speaker before the other speaker's delta from the same frame arrives, losing recent instructions at handoff.
## What changed
Use raw frame recency to trim timed frameless bidirectional transcripts within the 8 KiB budget. Defer eviction until the frame advances or the transcript is consumed, and retain both speakers' suffixes when they share the newest frame. Preserve conversation order and fall back to FIFO eviction when timing is unavailable.
Track trimmed prefixes so final transcripts can extend retained speech without making delayed finals appear newer.
## Testing
Add regression coverage for frame recency, same-frame event ordering, UTF-8 truncation, delayed finals, and untimed fallback. Add an app-server handoff test verifying that both speakers' recent speech and the latest user instruction survive while live transcript notifications remain unchanged.
GitOrigin-RevId: b01b729d8bdd4acb26e0385442ce40819355deac
## Why
Canceled classifier requests need bounded cleanup that preserves received token usage and returns only safe WebSocket connections to the pool.
## What changed
- Own classifier response streams and connection leases together, interrupting dropped responses with a one-second cleanup deadline.
- Require a matching interruption acknowledgement and response completion before reusing a canceled response's socket. Retire sockets on rejection, invalid acknowledgements, or timeout while preserving any received completion and token usage.
- Skip requests canceled before dispatch and return their unused sockets to the pool.
- Distinguish cancellation from ordinary steering, which continues draining without requiring an interruption acknowledgement.
## Testing
Add regression coverage for cancellation, acknowledgement ordering and failures, cleanup timeouts, socket reuse, and token accounting under backpressure. Extend steering tests to cover rejected and missing acknowledgements.
GitOrigin-RevId: c4e7aac49d36cea2c2d0e9f2ba5f6147f67471dd
## Why
Terminal `response.incomplete` errors discard reported token usage. Counting that usage as a new context baseline can also double-count input and trigger unnecessary compaction on the next turn.
## What changed
- Carry the response ID and valid reported usage with incomplete-response errors.
- Record usage for sampling and foreground compaction in cumulative totals, rollout records, budgets, and token-usage contributors without replacing the context baseline. Preserve the original terminal error, including when accounting exhausts the budget or cancellation races the error.
- Charge incomplete background compaction requests to the rollout budget without publishing foreground token usage.
- Ignore missing or malformed usage rather than estimating it.
## Testing
Extend coverage for SSE and WebSocket incomplete responses, rollout usage records, compaction accounting, deferred token-count events, and token-usage contributors. Add regressions for unnecessary compaction on the next turn and background compaction publishing foreground usage.
GitOrigin-RevId: bf1ee6df3912052ab7c19e1c6af8ad3dc4b07675
## Why
Incomplete responses such as `max_output_tokens` were classified as stream errors, allowing retries and compaction model fallback after the response had hit its output limit.
## What changed
Introduce `ResponseIncomplete` errors that preserve the server-provided reason and stop stream processing immediately. Make these errors non-retryable and stop compaction without falling back to the selected model. Record incomplete responses separately in sampler failure metrics.
## Testing
Add regression tests that verify incomplete responses surface a terminal error without retries over SSE and WebSocket, and that compaction stops without retrying or switching models for both local and remote compaction.
GitOrigin-RevId: 6e107d441b9bf03fa47ea47b88eeb94b71a9f729
## What changed
- Add the disabled-by-default `output_token_replay` feature to request `output.encrypted_content` from OpenAI providers.
- Preserve encrypted message and tool-call output, message and function-call status, and output-text `annotations` and `logprobs` through history and replay. Update the generated protocol schemas and SDK types.
- Strip output ciphertext from requests to other providers while retaining it in stored history, including after replay is disabled. Remove encrypted content from memory extraction inputs.
## Testing
Add integration coverage for the feature/provider combinations, metadata replay across follow-up requests, and history preservation when resuming with replay disabled.
GitOrigin-RevId: 77729a021b6771933402aebe74ece8eb39c057e2
## What changed
- Reformat Markdown tables and lists, login pages, inline visualization HTML, and sample JavaScript scripts.
- Escape union-type pipes in Python SDK API tables so they render within their cells.
- Keep the formatting-recipe test focused on the working directory and formatter commands by removing assertions on comment text.
GitOrigin-RevId: ae46d5b40fd013553de8b5ccf2c41d76be65aba8
## Why
Instant Connect can return a call ID before ICE activates the call, causing sideband attachment to briefly fail with HTTP `404` while the client connects media.
## What changed
Retry `404` responses from `connect_existing_call_sideband` up to three times, waiting 100, 300, and 800 milliseconds. Return the error once this activation retry budget is exhausted, and preserve the existing handling of other errors.
## Testing
Add WebSocket coverage for successful attachment after three `404` responses, failure after four `404` responses, and no retry on HTTP `410`. Verify that successful attachment does not overwrite the prepared session configuration.
GitOrigin-RevId: 5d18ff7eeeb9abd56388808828382f7350c7ae57
## Why
WebSocket rate-limit and overload errors ignored `Retry-After` advice and ended turns even when the server supplied a retry delay.
## What changed
Extract retry advice from nested `error.headers`, falling back to top-level `headers` when nested advice is missing or invalid. Use the shared HTTP header validation and retry-delay parser, and preserve the delay when mapping WebSocket errors.
## Testing
Add coverage for nested-header precedence, malformed headers, HTTP header validation, and recovery after an advised rate limit. Verify that overloads retry only while advice is present, without falling back to HTTP, and that headerless rate limits, usage limits, and `flex_unavailable` errors remain terminal.
GitOrigin-RevId: 61add0c36197e10402d6d3c325332aac72a5a521
## Why
Serial connection opens let a stalled handshake delay healthy sockets. Waiting until the idle pool is empty to refill also leaves classification bursts using HTTP fallback.
## What changed
- Open up to four WebSockets concurrently and start replenishing when the idle pool falls to half capacity. Preserve successful in-flight opens when another open fails or times out.
- Make explicit prewarming wait for an existing refill before checking readiness.
- Add `ResponsesWebsocketClient::connect_with_connector` and reuse one connector per sampler, initializing TLS configuration on a blocking task while retaining per-connection authentication and destination policy checks.
- Record connection lease durations for admission, setup, and total time, tagged by transport.
## Testing
Add regression tests for a stalled handshake alongside a healthy socket and replenishment before the last idle socket is leased. Update cooldown, authentication, and transport retry tests to account for concurrent handshake completion.
GitOrigin-RevId: f4a8258977c1f2a58e00c6badc7b599280ef22ad
## Why
Assistant answer text may be followed by more output or tools. A distinct `partial_answer` phase lets callers represent this separately from commentary and a terminal `final_answer`.
## What changed
- Add `MessagePhase::PartialAnswer` and expose `partial_answer` in the app-server schemas and generated TypeScript and Python types.
- Keep the TUI working indicator visible after partial answers and exclude them from the thread history's final-answer summary field.
- Route partial answers to the speakable channel in realtime V3 BEM-tag mode.
## Testing
Add coverage for phase preservation through serialization, input-message conversion, and SSE item start/completion events. Extend TUI snapshot coverage to verify that partial answers keep the working indicator visible while final answers hide it.
GitOrigin-RevId: 1ae3ad9264b6e36ae1cbe6b93df3d345ab4460e6
## What changed
- Carry opaque `review_target` values through misalignment errors and expose optional `reviewTarget` in the app-server protocol and generated clients. Preserve the value verbatim and redact it from debug output.
- Keep target-based `misalignment_override` metadata on turn requests, preserve it across steering updates, and exclude it from compaction and external MCP metadata. Reject overrides from global Responses metadata while retaining legacy timestamp behavior for per-turn overrides.
## Testing
Extend coverage for error propagation, debug and rollout redaction, steering preservation, and metadata filtering. Verify that automatic compaction omits the review target while turn requests before and after compaction retain it.
GitOrigin-RevId: 8c1abb076a206ff5927a4acc31bd25e3efc2bff9
## What changed
Move base instructions from the top-level `instructions` field into a prefixed `developer` message in `input` for standard Responses requests, matching Responses Lite. Remove the field from HTTP and WebSocket request types and assign instruction messages stable IDs derived from the thread and instruction text.
## Testing
Extend standard and Responses Lite tests to cover instruction message serialization, stable IDs, and resume consistency. Update WebSocket tests to verify that changed instructions trigger a full request without `previous_response_id`.
GitOrigin-RevId: e2de4b8e30a9339377fac133ceda981abd1a112b
## Why
Streamed `response.failed` events can carry retry advice in `response.error.headers`. Ignoring it loses server-provided delays or uses a conflicting delay from a rate-limit message.
## What changed
- Parse nested `Retry-After` headers using the existing HTTP header validation and retry parser, and preserve them for rate-limit, overload, and generic retryable failures.
- Prefer valid headers over rate-limit message advice, falling back to the message when the header is absent or invalid. Keep terminal error classifications unchanged.
- Reject unrepresentable plaintext retry durations without panicking.
## Testing
Add regression coverage for header precedence, validation, and terminal errors, plus retry-loop coverage for SSE sampling, remote compaction, and WebSocket failures.
GitOrigin-RevId: 46433c2e9e6d5e804211ef96950fadd70e322b49
## Why
Gateways that inspect request bodies incrementally need routing fields before potentially multi-megabyte prompts so they can route requests without buffering the input first.
## What changed
Serialize `model`, `stream`, and optional `service_tier` before `input` in both HTTP Responses requests and WebSocket `response.create` messages. Preserve field values and omission of unset `service_tier`.
## Testing
Add an HTTP regression test with a 2 MiB input that checks routing field order and unchanged JSON contents. Extend WebSocket serialization coverage to assert the routing fields appear first, and check that HTTP requests omit unset `service_tier`.
GitOrigin-RevId: dbc174e1fe7ca30d0c00f075bc37dd5f97b27b37
## Why
Responses overload and exhausted-request errors could terminate despite server retry advice, and WebSocket-to-HTTP fallback could issue a request before the advised deadline.
## What changed
- Allow `ServerOverloaded` and `RetryLimit` errors to retry when server advice is present, within the configured retry budgets.
- Preserve `Retry-After` headers from rejected WebSocket upgrades and wait for the server deadline before falling back to HTTP.
- Keep quota, usage-limit, and policy failures terminal even when they include retry advice.
## Testing
Add coverage for HTTP overload and rate-limit recovery, WebSocket upgrade rejection and fallback timing, retry deadline preservation through error mapping, and retry limits for sampling and manual and automatic compaction.
GitOrigin-RevId: 72d30a9f078652664754dfa971fd7dc7a78f20e8
## Why
Responses stopped by `content_filter` previously followed generic stream-error retry handling without guidance about the block. Retried requests need instructions to explain the limitation and offer a permitted alternative.
## What changed
- Recognize content-filter failures as a distinct error while preserving retryability and the existing public error representation.
- Append developer guidance to the conversation after each sampling block, retaining delivered output and using the existing retry handling.
- Support a model catalog override through `model_messages.content_filter_guidance`. Missing, null, blank, or values exceeding 512 UTF-8 bytes fall back to bundled guidance.
- The bundled guidance tells the model to avoid reproducing or working around blocked content, briefly explain the limitation, offer a permitted alternative, and continue unrelated authorized work.
## Testing
Add a scenario and request-history snapshot covering repeated blocks, catalog guidance on each retry, retained partial output, and eventual completion. Add error-mapping assertions for retryability, public error classification, and error text.
GitOrigin-RevId: ba0c201b3a5a3acf5cfa971332027240e0690225
Stop logging response headers on successful WebSocket connections and payload
previews for tool calls. Keep the connection URL, tool name, and thread ID in
their respective log entries.
GitOrigin-RevId: 1d410b6aef8f12153da9bfc60c515b8e797a9de0
## Why
Steering an active WebSocket response previously dropped the connection and
resent the full history. Draining the response preserves the connection and
allows the follow-up request to continue with `previous_response_id`.
## What changed
- Drain WebSocket responses when steering. For models using
`use_responses_lite`, first send `response.interrupt` with
`mode: "discard_partial_items"` once the response ID is available.
- Treat `response.incomplete` with reason `interrupted` as completion with
`end_turn: false`, preserving token usage and allowing the turn to continue.
Other incomplete reasons remain errors.
## Testing
Update the steering integration test to cover completed and discarded reasoning
items, asserting connection reuse, incremental follow-up input, and token usage
from the interrupted response.
GitOrigin-RevId: bc714b713e797cf1a67e3b26ffbf7664cb92eb33
## What changed
Move `response.failed` classification and rate-limit retry-delay parsing into `responses_error.rs`, preserving existing error mappings, fallback messages, and malformed-payload behavior. Replace the inline classification logic in the SSE event handler with `parse_failed_response`.
## Testing
Add regression coverage for missing or malformed failed-response payloads, including invalid values in unused error fields. Move the existing retry-delay parsing tests alongside the extracted parser.
GitOrigin-RevId: 6faa09b3fb312f1990509eef7088c8a086724615
Move JSON-to-HTTP header conversion from `responses_websocket` into the
crate-wide `responses_headers` module and update the WebSocket code to use it.
Preserve conversion of string, number, and boolean values, while continuing to
ignore invalid header names and unsupported or invalid values.
GitOrigin-RevId: f179de214cf763549b5b8a898dc4bcda3ce07ebd
## Why
Usage limit errors discarded the server-selected window responsible for the limit, leaving turn and compaction analytics unable to distinguish five-hour limits from weekly limits.
## What changed
Preserve optional `limit_window_minutes` from usage limit error responses in `UsageLimitReachedError` and report it as `usage_limit_window_minutes` in turn and compaction events. Accept unsigned integers that fit in `u16`; treat missing, null, malformed, or out-of-range values as unknown. Events for other error kinds report `null`.
## Testing
Add coverage for HTTP and wrapped WebSocket error mapping, analytics serialization and reduction, and app-server propagation to turn and local/remote compaction events. Cases include five-hour and weekly windows, missing or invalid values, and unrelated errors.
GitOrigin-RevId: a32066a007559d91a3c868a756bfd0f976777176
## What changed
- Recognize `flex_unavailable` in HTTP 429 responses and streamed `error` and `response.failed` events. End normal turns without retries and report `Flex capacity unavailable.`
- Expose `flexUnavailable` through the core and app-server protocols and generated schemas and SDK types.
- Keep Flex capacity failures eligible for Guardian review and sampler retries, with a distinct telemetry classification.
- Map unexpected HTTP responses to `HttpConnectionFailed`, preserving their status codes.
## Testing
Add regression coverage for terminal Flex failures over HTTP, SSE, and WebSocket, HTTP 504 status preservation after a stream retry, and protocol serialization. Update Guardian retry and sampler recovery tests to cover Flex failures.
GitOrigin-RevId: f2e03a6d2d36a6d9748c35952e40397f8670d121
## Why
Editing recent conversation images failed when the selected images included a file reference from a message or tool output.
## What changed
Use `ImageReference` in `ImageEditRequest` to accept both `image_url` and `file_id` inputs. Pass file references from conversation history through to image edit requests while preserving image selection order.
## Testing
Update tests to cover file references from messages and tool outputs, and serialization of mixed inline and file references in image edit requests.
GitOrigin-RevId: 2764bf36a81a62ddd25cf9df27346a434a10c449
## Why
File blob uploads previously failed immediately on HTTP 502 and 504 responses, even though these transient gateway errors can recover on retry.
## What changed
Treat HTTP 502 and 504 as retryable alongside 503, using the existing attempt limit, upload deadline, and server-provided retry delays.
## Testing
Expand upload retry tests to cover recovery from 502 and 504, mixed transient responses, exhausted attempts, and retry delays exceeding the remaining budget. Verify that 403, 409, and 500 responses are not retried, including responses without an Azure error code.
GitOrigin-RevId: a8e2f6bcf69338fb6d24f4c2d1ee333fb77bdbc0
cooper-oai