Keep the upstream payload prefixed for OAuth while passing the unprefixed request body into response translators. This avoids proxy_ leaking into OpenAI Responses echoed tool metadata while preserving the Claude OAuth workaround.
Prefix tool names with proxy_ for Claude OAuth requests and strip the prefix from streaming and non-streaming responses to restore client-facing names.
Updates the Claude executor to:
- add prefixing for tools, tool_choice, and tool_use messages when using OAuth tokens
- strip the prefix from tool_use events in SSE and non-streaming payloads
- add focused unit tests for prefix/strip helpers
Refactored `applyPayloadConfig` to `applyPayloadConfigWithRoot`, adding support for default rule validation against the original payload when available. Updated all executors to use `applyPayloadConfigWithRoot` and incorporate an optional original request payload for translations.
Anthropic API does not allow extended thinking when tool_choice is set
to "any" or a specific tool. This was causing 400 errors when using
features like Amp's /handoff command which forces tool_choice.
Added disableThinkingIfToolChoiceForced() that removes thinking config
when incompatible tool_choice is detected, applied to both streaming
and non-streaming paths.
Fixesrouter-for-me/CLIProxyAPI#630
Refined header assignment to use `x-api-key` for Anthropic API requests, ensuring correct authorization behavior based on request attributes and URL validation.
- Added support for parsing and normalizing dynamic thinking model suffixes.
- Centralized budget resolution across executors and payload helpers.
- Retired legacy Gemini-specific thinking handlers in favor of unified logic.
- Updated executors to use metadata-based thinking configuration.
- Added `ResolveOriginalModel` utility for resolving normalized upstream models using request metadata.
- Updated executors (Gemini, Codex, iFlow, OpenAI, Qwen) to incorporate upstream model resolution and substitute model values in payloads and request URLs.
- Ensured fallbacks handle cases with missing or malformed metadata to derive models robustly.
- Refactored upstream model resolution to dynamically incorporate metadata for selecting and normalizing models.
- Improved handling of thinking configurations and model overrides in executors.
- Removed hardcoded thinking model entries and migrated logic to metadata-based resolution.
- Updated payload mutations to always include the resolved model.
Fixes an issue where Claude thinking models would return 400 errors when
the thinking.budget_tokens was greater than or equal to max_tokens.
Changes:
- Add MaxCompletionTokens: 128000 to all Claude thinking model definitions
- Add ensureMaxTokensForThinking() function in claude_executor.go that:
- Checks if thinking is enabled with a budget_tokens value
- Looks up the model's MaxCompletionTokens from the registry
- Ensures max_tokens is set to at least the model's MaxCompletionTokens
- Falls back to budget_tokens + 4000 buffer if registry lookup fails
This ensures Anthropic API constraint (max_tokens > thinking.budget_tokens)
is always satisfied when using extended thinking features.
Fixes: #339🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
**feat(handlers, executor): add Gemini 3 Pro Preview support and refine Claude system instructions**
- Added support for the new "Gemini 3 Pro Preview" action in Gemini handlers, including detailed metadata and configuration.
- Removed redundant `cache_control` field from Claude system instructions for cleaner payload structure.
**fix(executor): replace redundant commented code with `checkSystemInstructions` helper**
- Replaced commented-out `sjson.SetRawBytes` lines with the new `checkSystemInstructions` function.
- Centralized system instruction handling for better code clarity and reuse.
- Ensured consistent logic for managing `system` field across Claude executor flows.
- Commented out multiple instances of `sjson.SetRawBytes` for setting `system` key to Claude instructions as they are redundant.
- Code cleanup to improve clarity and maintainability without affecting functionality.
- Add Claude thinking model definitions (sonnet-4-5-thinking, opus-4-5-thinking variants)
- Add Thinking support for antigravity models with -thinking suffix
- Add injectThinkingConfig() for automatic thinking budget based on model suffix
- Add resolveUpstreamModel() mappings for thinking variants to actual Claude models
- Add extractAndRemoveBetas() to convert betas array to anthropic-beta header
- Update applyClaudeHeaders() to merge custom betas from request body
Closes#324
Introduce `PayloadConfig` in the configuration to define default and override rules for modifying payload parameters. Implement `applyPayloadConfig` and `applyPayloadConfigWithRoot` to apply these rules across various executors, ensuring consistent parameter handling for different models and protocols. Update all relevant executors to utilize this functionality.
- Standardized User-Agent strings for Codex and Claude executors to improve request tracing and compatibility.
- Updated header insertion logic in both executors for consistency.
feat(runtime): add Brotli and Zstd compression support, improve response handling
- Implemented Brotli and Zstd decompression handling in `FileRequestLogger` and executor logic for enhanced compatibility.
- Added `decodeResponseBody` utility for streamlined multi-encoding support (Gzip, Deflate, Brotli, Zstd).
- Improved resource cleanup with composite readers for proper closure under all conditions.
- Updated dependencies in `go.mod` and `go.sum` to include Brotli and Zstd libraries.
- Introduced model alias mapping for Claude configurations, enabling upstream and client-facing model name associations.
- Added `computeClaudeModelsHash` to generate a consistent hash for model aliases.
- Implemented `normalizeClaudeKey` function to standardize input API key configuration, including models.
- Enhanced executor to resolve model aliases to upstream names dynamically.
- Updated documentation and configuration examples to reflect new model alias support.
- Updated the Execute methods in various executors (GeminiCLIExecutor, GeminiExecutor, IFlowExecutor, OpenAICompatExecutor, QwenExecutor) to return a response and error as named return values for improved clarity.
- Enhanced error handling by deferring failure tracking in usage reporters, ensuring that failures are reported correctly.
- Improved response body handling by ensuring proper closure and error logging for HTTP responses across all executors.
- Added failure tracking and reporting in the usage reporter to capture unsuccessful requests.
- Updated the usage logging structure to include a 'Failed' field for better tracking of request outcomes.
- Adjusted the logic in the RequestStatistics and Record methods to accommodate the new failure tracking mechanism.
- Added detailed logging of upstream request metadata including URL, method, headers, and body for Codex, Gemini, IFlow, OpenAI Compat, and Qwen executors.
- Implemented error logging for API response failures to capture errors during HTTP requests.
- Introduced structured logging for authentication details (AuthID, AuthLabel, AuthType, AuthValue) to improve traceability.
- Updated response logging to include status codes and headers for better debugging.
- Ensured that all executors consistently log API interactions to facilitate monitoring and troubleshooting.
When from==to (Claude→Claude scenario), directly forward SSE stream
line-by-line without invoking TranslateStream. This preserves the
multi-line SSE event structure (event:/data:/blank) and prevents
JSON parsing errors caused by event fragmentation.
Resolves: JSON parsing error when using Claude Code streaming responses
fix: correct SSE event formatting in Handler layer
Remove duplicate newline additions (\n\n) that were breaking SSE event format.
The Executor layer already provides properly formatted SSE chunks with correct
line endings, so the Handler should forward them as-is without modification.
Changes:
- Remove redundant \n\n addition after each chunk
- Add len(chunk) > 0 check before writing
- Format error messages as proper SSE events (event: error\ndata: {...}\n\n)
- Add chunkIdx counter for future debugging needs
This fixes JSON parsing errors caused by malformed SSE event streams.
fix: update comments for clarity in SSE event forwarding
- Added new "Gemini 2.5 Flash Image Preview" model definition, with enhanced image generation capabilities.
- Increased scanner buffer size to 20,971,520 bytes across executors and translators to handle larger payloads.
- Added `newProxyAwareHTTPClient` to centralize proxy configuration with priority on `auth.ProxyURL` and `cfg.ProxyURL`.
- Integrated enhanced proxy support across executors for HTTP, HTTPS, and SOCKS5 protocols.
- Refactored redundant HTTP client initialization to use `newProxyAwareHTTPClient` for consistent behavior.
- Deleted `ClientAdapter` implementation and associated fallback methods.
- Removed legacy executor logic from `codex`, `claude`, `gemini`, and `qwen` executors.
- Simplified `handlers` by eliminating `UnwrapError` handling and related dependencies.
- Cleaned up `model_registry` by removing logic associated with suspended clients.
- Updated `.gitignore` to ignore `.serena/` directory.
- Implemented `TokenCount` transform method across translators to calculate token usage.
- Integrated token counting logic into executor pipelines for Claude, Gemini, and CLI translators.
- Added corresponding API endpoints and handlers (`/messages/count_tokens`) for token usage retrieval.
- Enhanced translation registry to support `TokenCount` functionality alongside existing response types.
- Added `LoggerPlugin` to log usage metrics for observability.
- Introduced a new `Manager` to handle usage record queuing and plugin registration.
- Integrated new usage reporter and detailed metrics parsing into executors, covering providers like OpenAI, Codex, Claude, and Gemini.
- Improved token usage breakdown across streaming and non-streaming responses.
- Enhanced support for extracting system instructions from input arrays.
- Improved input message role and type determination logic for consistent message processing.
- Refined instruction handling logic across translator types for better compatibility.
- Enhanced support for extracting system instructions from input arrays.
- Improved input message role and type determination logic for consistent message processing.
- Refined instruction handling logic across translator types for better compatibility.
- Integrated ZSTD decompression via `github.com/klauspost/compress` for responses with "zstd" content-encoding.
- Added helper `hasZSTDEcoding` to detect ZSTD-encoded responses.
- Updated response handling logic to initialize and use a ZSTD decoder when necessary.
refactor(api-handlers): split streaming and non-streaming response handling
- Introduced `handleNonStreamingResponse` for processing non-streaming requests in `ClaudeCodeAPIHandler`.
- Improved code clarity by separating streaming and non-streaming logic.
fix(service): remove redundant token refresh interval assignment logic in `cliproxy` service.
- Introduced `CountTokens` method to Codex, Claude, Gemini, Qwen, OpenAI-compatible, and other executors.
- Implemented `ExecuteCount` in `AuthManager` for token counting via provider round-robin.
- Updated handlers to leverage `ExecuteCountWithAuthManager` for streamlined token counting.
- Added fallback and error handling logic for token counting requests.
- Introduced `EnsureHeader` in `internal/misc/header_utils.go` to streamline header setting across executors.
- Updated Codex, Claude, and Gemini executors to utilize `EnsureHeader` for consistent header application.
- Incorporated Gin context headers (if available) into request header manipulation for better integration.
- Added detailed debug logs in all executors (Codex, Claude, Gemini, Qwen, OpenAI-compatible) to capture HTTP status and response body for failed API requests.
- Added `last_refresh` timestamp to metadata for Codex, Claude, Qwen, and Gemini executors.
- Implemented `extractLastRefreshTimestamp` utility for parsing diverse timestamp formats in management handlers.
- Ensured consistent update and preservation of `last_refresh` in file-based auth handling.
- Introduced `logrus` for structured debugging across all executors.
- Added debug log messages in `Refresh` methods for better traceability.
- Updated `Manager` to log additional details during refresh checks.
