- Compact valid upstream JSON error response payloads to single-line formatting to ensure compatibility with SSE streams while preserving error data
Closes: #6434
An attachment a translator could not represent was dropped without a word, and
a turn whose only content was that attachment went upstream as an empty
messages, parts, or input array. The model then answered as if nothing was
attached, or the upstream rejected a request that never named the attachment.
A base64 document now reaches OpenAI as a file part. A file the target cannot
receive, such as container_upload, an OpenAI file_id, or input_audio sent to
Claude, is reported by name so the route answers 400 instead of forwarding an
empty turn. Text beside it still travels.
One rule, ErrIfNothingLeft, decides when to refuse, and RegisterCheckedRequest
carries the refusal on the request envelope. Every executor that translates
checks it before calling upstream. The exported converters keep their []byte
signature, so no caller or sibling test had to change.
- Reject speech-only models (such as `grok-tts` and `grok-voice-tts-1.0`) on non-speech entry protocols to prevent invalid chat upstream routing
- Mark upstream 404 errors for xAI speech as request-scoped unless indicating model unavailability, preventing unnecessary credential rotation and model cooldowns
- Align xAI speech payload finalizer protocol with `openai` for consistent payload override rules
- Track `finish_reason` presence across initial and streaming response chunks
- Return a terminal bad gateway error instead of `[DONE]` when upstream stream closes prematurely without a finish reason
Closes: #6371
- Treat Claude missing thread state 404 errors as client errors to bypass credential rotation and cooldown
- Inject `thread_not_found` error code into Claude error response details to signal client-side replay
- Support extracting upstream response bodies across error formatting and classification helpers
Closes: #6344
- Add `FlushResponse` to prefer error-aware flushing and unwrap wrapped response writers
- Bypass Gin's void `Flush` while maintaining header accounting via `WriteHeaderNow` to keep transport errors observable
Closes: #6332
- Retain `freeform` apply_patch tool type for native template models when `enable-apply-patch` is false
- Update client configuration documentation for `client.codex.enable-apply-patch`
Closes: #6286
Move shared Codex, Claude, and xAI provider options under upstream so
they apply consistently to OAuth and API-key credentials. Migrate
historical OAuth paths, preserve OAuth-only header scope, and expose the
new upstream management API path.
- Fall back to top-level token usage when Codex response usage contains only service tier metadata
- Propagate parsed service tier while retaining top-level token counts
- Avoid early return during stream usage observation for chunks with zero token counts
Closes: #6261
- Handle apply_patch custom tool call streaming deltas and snapshots for Responses API translation
- Track and resolve step indexes, function call identities, and error states across stream events
- Propagate apply_patch tool input errors and upstream failure states in stream and non-stream responses
- Return Bad Gateway error when Codex upstream stream closes before emitting any payload
- Allow failover to subsequent credentials on premature stream disconnect before output
- Map HTTP 408 Request Timeout status to server_error in base error response handlers
Closes: #6247
- Add `Path` field to model execution requests and plugin interceptor responses
- Propagate custom request paths through execution metadata
- Allow image models during execution when entry or exit protocol is `openai-image`
Closes: #6196
- Parse and pass `quality` in xAI image generation and edit requests
- Extract `quality` field from JSON payloads and multipart form data
- Include `quality` in the upstream xAI image request payload when provided
Closes: #6157
- Share request payload reference during session identity enrichment when original request is unset
- Avoid cloning request payload when interceptors do not mutate the body
- Pass request body directly to handler and conductor interceptors
Closes: #6101
- Marshal Codex client catalog as compact single-line JSON without HTML escaping
- Use compact base instructions for non-template models to stay within the 1MiB payload limit
- Keep `apply_patch_tool_type`, `upgrade`, and `availability_nux` present as null instead of deleting them
- Add `trusted-proxies` configuration option with IP and CIDR validation
- Apply trusted proxies to Gin engine to properly resolve forwarded client IPs
- Record `ResolvedClientIP` in request metadata and usage queue details
Closes: #6024
- Allow prewarm requests to be handled locally after an existing request turn
- Reset transcript root for mid-connection prewarms that omit `previous_response_id`
- Reuse local prewarm evaluation flag across request normalization and response handling
Closes: #6006
- Wait for bootstrap initialization before advancing mock clock in Codex WebSocket tests
- Reorder data channel closures in OpenAI responses stream error tests
- Drain stream chunks in byte-cap release test to prevent dangling goroutines
Closes: #5981
- Implement `streamCodexDuplex` to handle bidirectional WebSocket interactions for Codex
- Support steering continuations (`response.steer`) and queued creations (`response.create`, `response.append`)
- Preserve response settings and state transitions across steering updates and automatic successors
- Isolate connection-level failures with `codexDuplexConnectionError` to prevent unnecessary credential cooldowns
Closes: #5968
- Consolidate stream error code and error type derivations into paired error classes
- Map `StatusRequestTimeout` to `server_error` instead of `invalid_request_error` so interrupted streams remain retryable by clients
Closes: #5931
- Detect Devin models by ID prefix, type, ownership, model registry metadata, or provider.
- Append `(Devin)` to model display names in Codex client responses when not already present.
- Set model `type` to `devin` in home Codex model formatting when served by the Devin provider.
Grok Imagine already supports the 9:20 phone ratio, but the OpenAI image
handler whitelist dropped it and fell back to 1:1. Accept 9:20 from
aspect_ratio and from size, and keep unknown ratios on the existing
fallback.
- Introduce `authUnavailableError` with detailed recovery hints, status codes, and retry-after headers.
- Refactor scheduler logic to incorporate local retry deadlines and classify cooldown states without altering quota assessments.
- Refine error propagation to properly handle upstream errors and enrich context for retryable scenarios.
- Add comprehensive tests for retry deadlines, scheduler behavior, and error handling during credential unavailability.
- Ensure embedded recovery hints in HTTP headers for improved client-side handling and diagnostics.
Closes: #5842
- Add `ForcedProvider` and `AuthID` fields to `HostModelExecutionRequest` and `ModelExecutionRequest`.
- Pin credentials via `WithPinnedAuthID` and forward forced provider options during model execution and streaming.
- Forward forced provider and auth ID from plugin host callbacks to model execution handlers.
Closes: #5814
- Verify observed compaction response replay and retention across subsequent turns.
- Test plugin route overrides, provider auth pinning, and home runtime auth handling during compaction.
Closes: #5811
- Add `WriteModelListResponse` to `BaseAPIHandler` to apply plugin interceptors and record request lifecycles for model catalog responses.
- Update OpenAI, Claude, Gemini, Grok, and Codex model listing endpoints to route responses through the unified interceptor helper.
Closes: #5742
- Resolve canonical templates using metadata model IDs for model aliases and prefixed routes.
- Apply descriptions, base instructions, and thinking support overrides to matched templates.
- Restrict protocol capabilities and reasoning levels based on provider support.
Closes: #5699
- Track trailing carriage returns across chunk boundaries in `sseJSONValidationState`.
- Strip leading newline in subsequent chunks to prevent duplicate newline insertion from split CRLF sequences.
- Reset trailing carriage return state upon stream completion.
Closes: #5657
- Format streaming error payloads with nested error objects matching official OpenAI Responses SSE specifications.
- Extract and propagate sequence numbers from upstream terminal events and framer states.
- Use `json.Number` to prevent precision loss for large integers and token metrics.
- Sanitize sensitive keys recursively across nested error objects without dropping custom fields.
- Track pending synthetic prewarm response IDs to merge warmup inputs into subsequent delta followups.
- Normalize transcript replacements when followups do not reference the prewarm parent response ID.
- Validate that the `input` field is an array for `response.create` requests.
- Allow `function_call_output` items without a `call_id` when a non-empty tool name is present.
Closes: #5631
- Register builtin model definitions for `gpt-image-2.5`, `gpt-image-2.5-flare`, and `gpt-image-2.5-sunburst`.
- Update OpenAI image handlers and request routing to recognize GPT Image 2.5 models.
- Support direct image generation and edit execution for GPT Image 2.5 variants in the Codex executor.
- Apply client visibility overrides to hide new builtin image models where appropriate.
- Introduce `IsTerminalAuthError` and `NewTerminalAuthError` to identify permanent upstream authentication failures.
- Return terminal auth errors from candidate selection and scheduling when all available credentials fail with unauthorized errors.
- Support `BuildErrorResponseBodyWithError` to format terminal upstream auth errors as non-retryable `upstream_authentication_required` responses.
- Propagate terminal error classifications and the `retryable` field across HTTP and WebSocket response handlers.
Closes: #5645
- Initialize `util.SessionIDResolver` to resolve session IDs from request contexts, metadata, and headers.
- Ensure canonical session metadata is injected into execution options across execution flows.
- Propagate session context in executors to support `$CPA-SESSION-ID` expansion in custom headers.
- Sync cleared or updated session identities back to execution contexts during conductor execution.
Closes: #5690
- Reuse coresession.ExtractSessionInfo across HTTP headers and request payloads to unify canonical session prefix namespaces with the scheduler.
- Extract hierarchical session identities in two phases: initial extraction from request headers on entry, and authoritative deep extraction once request payloads and metadata are available.
- Support Claude Code multi-level subagents (X-Claude-Code-Agent-Id, metadata.agent_id) and Codex thread fork lineages.
- Propagate SessionID and ParentSessionID across ClientRequestMetadata, UsageReporter, and coreusage.Record without root_session_id.
- Include session_id and parent_session_id in queuedUsageDetail for Home LPushUsage forwarding and Redis consumption with self-loop guards.
- Add comprehensive test coverage for canonical headers, body extraction, ghost parent elimination, and self-referential loop guards.