Commit Graph
457 Commits
Author SHA1 Message Date
RCD 377697331b feat: sync with upstream v6.7.17-18
- Updated gemini_schema.go with improved schema cleaning
- Refined signature caching logic (kept 1h TTL for memory efficiency)
- Enhanced Claude/Gemini translator tests
- Fixed Gemini function-calling INVALID_ARGUMENT via relaxed tool validation
2026-01-23 19:26:02 +07:00
hkfires ecc850bfb7 feat(executor): apply payload rules using requested model 2026-01-23 16:38:41 +08:00
hkfires 7ca045d8b9 fix(executor): adjust model-specific request payload 2026-01-22 20:28:08 +08:00
hkfires abfca6aab2 refactor(util): reorder gemini schema cleaner helpers 2026-01-22 18:38:48 +08:00
sowar1987 a2f8f59192 Fix Gemini function-calling INVALID_ARGUMENT by relaxing Gemini tool validation and cleaning schema 2026-01-22 17:11:07 +08:00
XYenon 8c7c446f33 fix(gemini): preserve displayName and description in models list
Previously GeminiModels handler unconditionally overwrote displayName
and description with the model name, losing the original values defined
in model definitions (e.g., 'Gemini 3 Pro Preview').

Now only set these fields as fallback when they are missing or empty.
2026-01-22 15:19:27 +08:00
RCD e86d2b86d5 feat: sync with upstream v6.7.16
Major changes from upstream v6.7.16:
- Remove sessionID from signature caching API entirely
  - CacheSignature(modelName, text, sig) instead of 4 params
  - GetCachedSignature(modelName, text) instead of 3 params
  - HasValidSignature(modelName, sig) instead of 1 param
- Add Gemini fallback: returns "skip_thought_signature_validator" on cache miss
- Add sliding expiration (TTL refresh on cache access)
- Add GetModelGroup() for grouping cache keys by model family
- Simplify antigravity executor by removing project switch logic
- Fix OpenAI Responses request handler to skip empty assistant text
- Fix OpenAI Responses request to validate full JSON before parsing

Local fixes for compatibility:
- Add Embed stub method to AntigravityExecutor for interface compliance
- Update test to clear cache for isolation (ThinkingBlockWithoutSignature)
- Update Gemini test to expect thinking blocks to be annotated not removed
- Fix gjson.Valid check before treating tool output as raw JSON
2026-01-21 15:05:05 +07:00
hkfires 6dcbbf64c3 fix(executor): only strip maxOutputTokens for non-claude models 2026-01-21 10:49:20 +08:00
bexcodex 9f364441e8 Fix antigravity malformed_function_call 2026-01-20 19:54:54 +08:00
RCD e0fa10b669 feat: sync with upstream v6.7.13
Features ported from upstream:
- feat(claude): add native request cloaking support
- feat(translator): add candidate count (n param) support for Gemini
- fix(translator): consolidate multiple system messages into single message (#1064)
- feat(codex): add config toggle for codex instructions injection
- feat(auth): add hashed account ID to credential filenames for team plans
- feat(config): add support for raw JSON payload rules (default-raw, override-raw)
- feat(codex): add OpenCode instructions based on user agent

Also includes minor whitespace cleanup and code organization.
2026-01-20 14:08:05 +07:00
RCD 4a0f8232d5 refactor(claude): remove redundant system instructions check
The checkSystemInstructions was being called on every request but the
cloaking/system prompt handling is now done elsewhere. This removes
redundant processing.

Cherry-picked from CLIProxyAPI: 7831cba9
2026-01-20 11:49:21 +07:00
RCD 02b0714e3a feat(vertex): add Imagen image generation model support
Add support for Imagen 3.0 and 4.0 image generation models in Vertex AI:

- Add 5 Imagen model definitions (4.0, 4.0-ultra, 4.0-fast, 3.0, 3.0-fast)
- Implement :predict action routing for Imagen models
- Convert Imagen request/response format to match Gemini structure
- Transform prompts to Imagen's instances/parameters format
- Convert base64 image responses to Gemini-compatible inline data

Cherry-picked from CLIProxyAPI: 8734d4cb
2026-01-20 11:31:07 +07:00
RCD e6e7e6db30 feat(registry): support provider-specific model info lookup
Add InfoByProvider map to ModelRegistration to store provider-specific
ModelInfo, enabling different capabilities for the same model across
providers (e.g., direct API vs OpenRouter).

Changes:
- Add InfoByProvider field to ModelRegistration struct
- Update GetModelInfo to accept provider parameter for provider-specific lookup
- Update LookupModelInfo to accept variadic provider parameter
- Add LookupStaticModelInfo for fallback to static model definitions
- Update ApplyThinking to pass providerKey for registry lookups
- Update all GetModelInfo callers to use new signature
- Remove ensureMaxTokensForThinking test (function moved to thinking package)

Ported from CLIProxyAPI e641fde2
2026-01-20 11:15:39 +07:00
RCD 5dd078ffb3 refactor(claude): move max_tokens constraint enforcement to Apply method
Move ensureMaxTokensForThinking logic from claude_executor to
thinking/provider/claude/apply.go. This centralizes Claude-specific
thinking budget constraints in the thinking package.

Cherry-picked from c421d653.
2026-01-20 10:53:37 +07:00
RCD 8d8a010577 feat(executor): enhance payload translation with original request context
Add 6th parameter 'original' to applyPayloadConfigWithRoot to support
checking defaults against the original payload. This enables proper
default rule validation when translating between formats.

Based on upstream 2a663d5c.
2026-01-20 10:49:30 +07:00
RCD 5a38b6bc04 fix(executor): stop rewriting thinkingLevel for gemini
Remove automatic thinkingLevel -> thinkingBudget:-1 conversion
for non-gemini-3 models in antigravity executor.

Cherry-picked from 1d2fe553.
2026-01-20 10:40:07 +07:00
RCD f1d05cc027 feat(registry): expand antigravity model config
- Remove static Name mapping from AntigravityModelConfig
- Add new model entries: claude-sonnet-4-5, gpt-oss-120b-medium, tab_flash_lite_preview
- Add model skip list for unlisted models
- Update thinking limits (Max: 128000, ZeroAllowed: true)

Cherry-picked from c175821c.
2026-01-20 10:39:00 +07:00
Luis Pater 7831cba9f6 refactor(claude): remove redundant system instructions check in Claude executor 2026-01-20 11:02:52 +08:00
Luis Pater e02b2d58d5 Merge pull request #868 2026-01-20 10:57:24 +08:00
hkfires e641fde25c feat(registry): support provider-specific model info lookup 2026-01-20 10:01:17 +08:00
dinhkarate 8734d4cb90 feat(vertex): add Imagen image generation model support
Add support for Imagen 3.0 and 4.0 image generation models in Vertex AI:

- Add 5 Imagen model definitions (4.0, 4.0-ultra, 4.0-fast, 3.0, 3.0-fast)
- Implement :predict action routing for Imagen models
- Convert Imagen request/response format to match Gemini structure like gemini-3-pro-image
- Transform prompts to Imagen's instances/parameters format
- Convert base64 image responses to Gemini-compatible inline data
2026-01-20 01:26:37 +07:00
Luis Pater 5baa753539 Merge pull request #1099 from router-for-me/claude
refactor(claude): move max_tokens constraint enforcement to Apply method
2026-01-19 20:55:59 +08:00
hkfires 1d2fe55310 fix(executor): stop rewriting thinkingLevel for gemini 2026-01-19 19:49:39 +08:00
hkfires c175821cc4 feat(registry): expand antigravity model config
Remove static Name mapping and add entries for claude-sonnet-4-5,
tab_flash_lite_preview, and gpt-oss-120b-medium configs
2026-01-19 19:32:00 +08:00
hkfires c421d653e7 refactor(claude): move max_tokens constraint enforcement to Apply method 2026-01-19 15:50:35 +08:00
hkfires c7e8830a56 refactor(thinking): pass source and target formats to ApplyThinking for cross-format validation
Update ApplyThinking signature to accept fromFormat and toFormat parameters
instead of a single provider string. This enables:

- Proper level-to-budget conversion when source is level-based (openai/codex)
  and target is budget-based (gemini/claude)
- Strict budget range validation when source and target formats match
- Level clamping to nearest supported level for cross-format requests
- Format alias resolution in SDK translator registry for codex/openai-response

Also adds ErrBudgetOutOfRange error code and improves iflow config extraction
to fall back to openai format when iflow-specific config is not present.
2026-01-18 10:30:15 +08:00
Luis Pater bc7167e9fe feat(runtime): add model alias support and enhance payload rule matching
- Introduced `payloadModelAliases` and `payloadModelCandidates` functions to support model aliases for improved flexibility.
- Updated rule matching logic to handle multiple model candidates.
- Refactored variable naming in executor to improve code clarity and consistency.
2026-01-17 05:05:24 +08:00
Luis Pater 6600d58ba2 feat(codex): enhance input transformation and remove unused safety_identifier field
- Added logic to transform `inputResults` into structured JSON for improved processing.
- Removed redundant `safety_identifier` field in executor payload to streamline requests.
2026-01-16 19:59:01 +08:00
RCD f2eabe57c5 fix(codex): ensure instructions field exists in request body
Add check to ensure 'instructions' field exists in Execute, ExecuteStream,
and CountTokens functions before sending request to Codex API.

Cherry-picked from:
- CLIProxyAPI v6.7.4 (902bea2)
2026-01-16 15:25:50 +07:00
RCD a2d701b205 fix(executor): update gemini-3-pro model name check to use internal name
Change conditional check from gemini-3-pro-preview (alias) to
gemini-3-pro-high (internal name) for system instruction injection.

Cherry-picked from:
- CLIProxyAPI v6.7.3 (c3ef46f)
2026-01-16 15:25:50 +07:00
RCD 7b22c4fd4d feat(auth): cherry-pick CLIProxyAPI v6.6.108-109 core updates
- Add ensureAntigravityProjectID to fetch project_id during token refresh
- Simplify filestore Save() to use os.O_TRUNC instead of temp file rename
- Auto-fetch project_id for antigravity provider in readAuthFile()
- Add metadataEqualIgnoringTimestamps to prevent refresh loops
- Use reflect.DeepEqual for cleaner metadata comparison

Cherry-picked from:
- CLIProxyAPI v6.6.108 (086eb3d)
- CLIProxyAPI v6.6.109-dev (7248f65)
2026-01-16 15:25:49 +07:00
hkfires 902bea24b4 fix(codex): ensure instructions field exists 2026-01-16 15:38:10 +08:00
hkfires c3ef46f409 feat(config): supplement missing default aliases during antigravity migration 2026-01-16 13:37:46 +08:00
Luis Pater f571b1deb0 feat(config): add support for raw JSON payload rules
- Introduced `default-raw` and `override-raw` rules to handle raw JSON values.
- Enhanced `PayloadConfig` to validate and sanitize raw JSON payload rules.
- Updated executor logic to apply `default-raw` and `override-raw` rules.
- Extended example YAML to include usage of raw JSON rules.
2026-01-16 08:15:28 +08:00
Luis Pater 67f8732683 Merge pull request #1033 from router-for-me/reasoning
Refactor thinking
2026-01-15 20:33:13 +08:00
hkfires 4ad6189487 refactor(thinking): extract antigravity logic into a dedicated provider 2026-01-15 19:08:22 +08:00
Luis Pater 086eb3df7a refactor(auth): simplify file handling logic and remove redundant comparison functions
feat(auth): fetch and update Antigravity project ID from metadata during filestore operations

- Added support to retrieve and update `project_id` using the access token if missing in metadata.
- Integrated HTTP client to fetch project ID dynamically.
- Enhanced metadata persistence logic.
2026-01-15 13:29:14 +08:00
hkfires 5a77b7728e refactor(thinking): improve budget clamping and logging with provider/model context 2026-01-15 13:06:41 +08:00
hkfires ed8b0f25ee fix(thinking): use LookupModelInfo for model data 2026-01-15 13:06:41 +08:00
hkfires 6e4a602c60 fix(thinking): map reasoning_effort to thinkingConfig 2026-01-15 13:06:40 +08:00
hkfires 7f1b2b3f6e fix(thinking): improve model lookup and validation 2026-01-15 13:06:40 +08:00
hkfires a75fb6af90 refactor(antigravity): remove hardcoded model aliases 2026-01-15 13:06:39 +08:00
hkfires 72f2125668 fix(executor): properly handle thinking application errors 2026-01-15 13:06:39 +08:00
hkfires 0b06d637e7 refactor: improve thinking logic 2026-01-15 13:06:39 +08:00
Luis Pater 94e979865e Fixed: #897
refactor(executor): remove `prompt_cache_retention` from request payloads
2026-01-12 10:46:47 +08:00
hkfires 70a82d80ac fix(codex): only override instructions in responses for OpenCode UA 2026-01-11 15:19:37 +08:00
hkfires ac626111ac feat(codex): add OpenCode instructions based on user agent 2026-01-11 13:36:35 +08:00
Finessse 9f5eb93f64 sync(upstream): add HttpRequest support across executors
Cherry-picked from e8e3bc86. Adds HttpRequest method to ProviderExecutor
interface for injecting credentials into arbitrary HTTP requests.

- Added HttpRequest to conductor.go interface
- Added HttpRequest stubs to kiro, minimax, zhipu executors
- Added http-request example
- Updated mock executors in tests

Also kept our Embed method and CountTokens implementations.
2026-01-11 06:32:53 +07:00
Finessse 01e1d5aed2 sync(upstream): auth rotation atomic fix + memory leak fixes
- fix(auth): make provider rotation atomic (keep shouldKeepProviderOrder)
- fix(server): resolve memory leaks in failed attempts and codex cache
  - Add background cleanup goroutine for stale IP entries
  - Add cache expiry cleanup for codex prompt cache
- Update provider rotation tests for atomic rotation

Cherry-picked: 8be06255, 47dacce6
Merged with ProxyPilot's shouldKeepProviderOrder for gemini-3 fallback
2026-01-11 06:11:37 +07:00
Finessse 19bf935a2e sync(upstream): Gemini 3 Pro and Antigravity executor improvements
- feat(executor): enhance Antigravity payload with user role and dynamic system instructions
- feat(executor): add model-specific support for "gemini-3-pro" in execution
- feat(executor): update system instruction handling for Claude and Gemini-3-Pro models
- fix(executor): update gemini model identifier to gemini-3-pro-preview
- feat(executor): centralize systemInstruction handling

Cherry-picked: 67985d82, bda04eed, 1b2f9076, b6a0f7a0, 59a448b6
2026-01-11 06:03:27 +07:00