components/tool/mcp/v0.0.9

What's Changed

• feat(ark): export GetCacheExpiration by @meguminnnnnnnnn in #574
• feat(acl/openai): feat: add RequestPayloadModifier and ResponseMessageModifier by @JonXSnow in #570
• fix(openai): fix ResponseMessageModifier usage by @hi-pender in #575
• fix: nil interface issue in NewClient HTTPClient handling by @ComingCL in #578
• fix(ark): response meta's finish reason use status.String() & toolCall's type use FunctionToolCall.Type.String() by @JonXSnow in #580
• feat(deepseek): support thinking chain for tool_call by @hi-pender in #583
• feat(ark): extend generate_with_image example to use chat template by @shentongmartin in #584
• feat(acl/openai): when reasoning content is empty, try get reasoning content from delta extra by @JonXSnow in #582
• fix(volc_knowledge): invalid request json by @N3kox in #586
• feat(doc/readme): Add README_zh.md & update docs by @JonXSnow in #585
• fix(gemini): Merge Tool Messages into Content Parts by @yesiyu403 in #565
• refactor(acl/openai): remove word OpenAI in error msg by @hi-pender in #589
• feat(retriever): 在Segment结构体中添加sign_content字段并优先使用 by @regend in #569
• feat(ark): support reasoning content pass back to model for chat completions api by @JonXSnow in #590
• feat(acl/openai): when reasoning content is empty, try get reasoning content from chat completions message extra by @JonXSnow in #591
• feat(devops): fix non-constant format string for Errorf by @hi-pender in #593
• feat(acl/openai): use populateReasoningContentFromExtraFields to get the reason content from extraFields by @JonXSnow in #592
• fix(gemini): Set message role to "schema.User" for "user" role, and to "schema.Assistant" for other roles. by @JonXSnow in #594
• feat(ark): support ark model batch_chat by @gyhbbb in #596
• feat(openrouter): add openrouter chat model by @JonXSnow in #576
• feat(openrouter): Downgrade the go version to 1.18 by @JonXSnow in #603
• fix(ollama): data race in ut by @shentongmartin in #604
• feat(ark): ark response api support file type by @JonXSnow in #595
• feat(gemini): add reasoning token count by @JonXSnow in #608
• feat(ark): add reasoning token count by @JonXSnow in #609
• feat(gemini): add reasoning token count for callback output by @JonXSnow in #610
• feat(acl/openai): add reasoning token count by @JonXSnow in #611
• feat(qwen&openai): upgrade acl/openai to support reason tokens by @JonXSnow in #612
• feat(components/model): upgrade eino and acl/openai to latest by @JonXSnow in #613
• feat(ark): ark chat completions support max_completion_tokens by @JonXSnow in #597
• feat: support get span context from cozeloop callback; tool node add tool_call_id tag by @JqRrt in #600
• feat(gemini): use empty judgment logic instead of nil judgment by @JonXSnow in #617
• feat(opensearch2): add OpenSearch2 indexer and retriever components by @kaijchen in #601
• fix(es8): correct dense vector similarity parameter name by @kaijchen in #605
• feat(es7): add Elasticsearch 7 retriever and indexer components by @kaijchen in #606
• feat(es8): refine examples and documentation for ES8 retriever/indexer by @kaijchen in #607
• feat(opensearch3): add OpenSearch3 indexer and retriever components by @kaijchen in #614
• feat(ark): report model name in callback extra by @meguminnnnnnnnn in #627
• feat(model): support multicontent tool result by @meguminnnnnnnnn in #623
• feat(claude): add Google Vertex AI support by @shanemcd in #625
• feat: support gemini internal tool by @meguminnnnnnnnn in #629
• feat(components/model): Unify and enhance tool_choice and allowed_tools handling across models by @JonXSnow in #626
• feat(model/ark): support web search & add NewResponseAPIChatModel method by @JonXSnow in #632
• feat(model/ark): rename Enable Tool Web Search to Web Search by @JonXSnow in #637
• feat(gemini): export thought signature getters by @hi-pender in #636
• feat(components/model): upgrade acl/openai by @JonXSnow in #639
• docs(cloudsearch): improve README of ES and OpenSearch by @kaijchen in #631
• feat(gemini): support multi-modal tool result by @meguminnnnnnnnn in #644
• feat(model/gemini): add image config for gemini chat model by @JonXSnow in #646
• feat(model/ark): Refactor callback handling for Responses API by @JonXSnow in #649
• feat(es9): add Elasticsearch 9 indexer and retriever components by @kaijchen in #640
• fix(model/gemini): Correctly attach metadata to video input part by @JonXSnow in #650
• feat(milvus2): add milvus2 indexer and retriever components by @kaijchen in #628
• fix(model/ark):Implement the IsCallbacksEnabled method in the Ark Responses API by @JonXSnow in #653
• fix(gemini): prevent duplicate tool call IDs for same function calls by @shentongmartin in #652
• feat(openai): conv tool calls if multi content exists by @meguminnnnnnnnn in #656
• feat(openai): upgrade acl by @meguminnnnnnnnn in #657
• refactor(gemini): replace counter-based tool call IDs with UUIDs by @shentongmartin in #659
• fix(claude): Don't spam "MultiContent is deprecated" when it actually isn't used by @mikeb26 in #654
• feat(model/qwen): Add support for custom request headers and fields by @JonXSnow in #663
• feat(es/os): add IndexSpec to support automatic index creation by @kaijchen in #661
• feat(claude): add configurable, additional fields to request header and body by @phagenlocher in #675
• feat(cozeloop) tool message add name & split multi_content & support gemin… by @JqRrt in #664
• feat(ark): support partial assistant message by @meguminnnnnnnnn in #670
• feat(ci): pin typos version by @meguminnnnnnnnn in #683
• feat(adk/backend): implement Agentkit Sandbox and Local filesystem backend adapters by @JonXSnow in #651
• fix(adk): ensure streaming shell response is sent even when command produces no output by @JonXSnow in #689
• feat: support custom resource attributes in apmplus by @danielqsj in #690
• refactor: export backend struct types for external access by @JonXSnow in #697
• docs: optimize README files for all modules with examples by @shentongmartin in #702
• refactor(ark): remove explicit tool choice assignment from chat model by @JonXSnow in #695
• docs(gemini): 为gemini/examples添加了自定义URL的使用示例 by @SKDG042 in #442
• fix(gemini): gemini tool_call part position by @N3kox in #715
• feat(backend): enhance backend with ripgrep support, background execution and code optimizations by @JonXSnow in #707
• feat(cozeloop): add ADK agent support for tracing by @shentongmartin in #705
• feat(backend): allow Write method to overwrite existing files by @JonXSnow in #720
• fix: add missing readmes by @shentongmartin in #721
• fix(openrouter): register []*reasoningDetails with gob for checkpoint serialization by @jesse-engineer in #723
• fix(openai_acl): set lost reasoning_content for the assistant message by @N3kox in #724
• fix: missing reasoning content caused by libs/acl/openai by @N3kox in #725
• refactor(filesystem): remove formatPath helper and simplify path normalization by @JonXSnow in #726
• feat: add Agent Skills for Eino framework by @hi-pender in #729
• fix(agentkit): fix glob output parsing to use JSON array by @JonXSnow in #731
• feat(claude): add cache TTL support and separate option types for auto cache and manual breakpoint by @JonXSnow in #727
• fix(local): replace bufio.Scanner with bufio.Reader, implement Execute, and merge stderr to stdout in code template by @JonXSnow in #735
• fix: change pdf parser dependency by @xuwuruoshui in #630
• fix(libs): bump go-openai to v0.1.2 to preserve content field for tool messages with empty content by @shentongmartin in #753
• fix(libs): bump libs/acl/openai to v0.1.15 in dependent modules by @shentongmartin in #754
• fix(claude): align direct anthropic auth precedence with sdk behavior by @hi-pender in #759
• feat(model): support tool role in multi-content messages by @JonXSnow in #764
• fix(libs): bump libs/acl/openai to v0.1.16 in dependent modules by @JonXSnow in #765
• feat(gemini): support gemini multi text tool result by @meguminnnnnnnnn in #768
• feat(openrouter): add top-level cache control and support non-text parts by @JonXSnow in #766
• feat(openrouter): add signature field to reasoningDetails struct by @yesiyu403 in #772
• Feat openrouter max tokens by @yesiyu403 in #773
• fix(ark): make cache status tri-state to avoid sending caching field when unset by @JonXSnow in #771
• fix(ark): extract generic int64 extraction helper for message extra by @mrh997 in #776
• fix(openrouter): enhance reasoning details extraction by @yesiyu403 in #774
• [fix] cozeloop fix interrupt error by @JqRrt in #775
• fix(openai): use json.Unmarshal for reasoning content extraction by @JonXSnow in #778
• feat(components): upgrade acl/openai by @JonXSnow in #779
• feat(devops)!: add WithDevServerIP and change default listen to 127.0.0.1 by @JonXSnow in #789
• fix(cozeloop): agent span type by @JqRrt in #792
• feat(ark): add EnableReasoningContentPassback config for ResponsesAPI by @JonXSnow in #791
• feat(deepseek): support text-only multi content concatenation by @JonXSnow in #794
• feat(deepseek): add ThinkingConfig support for thinking/non-thinking mode by @JonXSnow in #798
• feat(deepseek): support ExtraFields via request-level option by @jesse-engineer in #799
• fix(claude): compatibility handling empty message by @meguminnnnnnnnn in #818
• feat(ark): support audio input in Responses API by @JonXSnow in #819
• fix(deepseek): propagate Thinking config in stream request by @JonXSnow in #823
• feat(ark): remove ResponseFormat limit for responses api by @mrh997 in #824
• feat(cozeloop): trace add agent_run_id by @JqRrt in #815
• feat(agentkit/local): add MultiModalRead for images and PDFs by @JonXSnow in #812
• fix(gemini): include ThoughtsTokenCount in CompletionTokens by @jesse-engineer in #827
• feat(libs/acl/langfuse): usageDetails schema and EndTrace by @jesse-engineer in #828
• feat(callbacks/langfuse): ResponseMeta usage mapping and trace fixes (depends on #828) by @jesse-engineer in #829
• feat: merge alpha/09 to main by @shentongmartin in #836
• feat(agenticgemini): support image config by @tangchaojun-bytedance in #840
• feat(gemini): support response logprobs by @JonXSnow in #841
• feat(acp): backend improvements by @fanlv in #842
• feat(agentic): support cache control for agenticclaude, agenticark, agenticopenai, and agenticgemini by @mrh997 in #839
• refactor(agenticopenai): use upstream prompt cache retention type directly by @mrh997 in #844
• chore: update cloudwego/eino dependency to v0.9.0 by @shentongmartin in #847
• feat(acp): support preserve tool call stream by @meguminnnnnnnnn in #850
• feat(agenticgemini): support server tools options by @tangchaojun-bytedance in #851
• refactor(qdrant): enhance embedding callback context and error wrapping by @sheep44044 in #845
• refactor(agenticopenai): flatten chat/ and responses/ into root package by @mrh997 in #849
• refactor(agentic-models): align Gemini API and tool config by @mrh997 in #852
• feat(ark): support ark embedding dimensions by @N3kox in #853
• chore: bump cloudwego/eino dependency from v0.9.0 to v0.9.1 by @shentongmartin in #854
• fix(agentic-models): fix ToolSearchCall mapstructure tag and add mapstructure support by @mrh997 in #855
• feat(document): add basic endpoint configuration for the S3 loader by @kklinan in #281
• fix(openai): merge ExtraFields in WithExtraFields instead of overwriting by @wucm667 in #793
• fix(ollama): include items structure in tool parameter definitions by @wucm667 in #797
• index option for indexer and retriver, milvius id' type, fix license by @tangchaojun-bytedance in #859
• fix(ark): add ProjectName in EmbeddingConfig to support preset endpoints by @Jmmt-mingrui in #831
• fix(deepseek): enable token usage in streaming responses by @wucm667 in #809
• fix(es8): adapt retriever to latest elastic typed API by @kexyseeds in #860
• docs: character adjustment README_zh.md by @shaxiaozz in #722
• optimize(openai): enhance APIError struct with InnerError field by @zbysir in #752
• fix(devops): correct CompletionTokens assignment typo in callback handler by @SKDG042 in #784
• fix(langfuse): correct nil check variable typo in extractModelInput by @SKDG042 in #785
• feat(readme): improve readme by @mrh997 in #870
• fix(claude): report full token usage for streaming responses by @mrh997 in #871
• test(deepseek): add unit tests for concatTextParts by @JonXSnow in #795
• refactor(adk): replace go-fitz with go-pdfium (WASM) for MultiModalRead by @JonXSnow in #869
• fix(claude): merge consecutive tool messages into single UserMessage by @mjlzz in #872
• fix(agentic): tolerate unknown stream variants instead of failing by @mrh997 in #875
• fix(agentic): merge adjacent tool-result messages for claude and gemini by @mrh997 in #876
• feat(agentic): support openai.ReasoningExtension for reasoning text by @mrh997 in #878
• fix(agenticclaude): align direct anthropic auth precedence with sdk behavior by @hi-pender in #884
• fix(semantic): correct split by @meguminnnnnnnnn in #887
• fix(callbacks): fix cozeloop token usage by @shentongmartin in #897
• feat(claude): add structured output support by @marwan38 in #709
• docs(agentic): enrich cache section with mermaid flowchart and examples by @mrh997 in #898
• chore(openai): enrich reasoning effort enums by @N3kox in #899
• feat(claude): add adaptive thinking support by @jesse-engineer in #902
• feat(libs/acl/langfuse): expose max event size option by @jesse-engineer in #910
• feat(callbacks/langfuse): expose max event size config by @jesse-engineer in #912
• feat(model): support mid-conversation system messages by @mrh997 in #914
• feat(claude): support Vertex service account JSON credentials by @shaikustin in #888
• feat(claude): add RequestTimeout config and per-request option by @mrh997 in #920
• fix(deepseek): map completion_tokens_details.reasoning_tokens in usage by @jesse-engineer in #921
• fix(claude): map output_tokens_details.thinking_tokens in usage by @jesse-engineer in #927
• feat(officialmcp): reconnecting session via ClientSession anti-corruption interface by @hi-pender in #889
• feat(claude): expose cache_creation_input_tokens in message.Extra and add Config.AutoCacheControl by @sin-bufan in #929
• fix(ark): support fast service tier in responses API by @Abtiotm in #938
• feat(officialmcp): support custom transport factories by @fanlv in #946
• fix(tool_mcp): fix InvokableRun crash on empty arguments #923 by @kayoch1n in #924

New Contributors

• @ComingCL made their first contribution in #578
• @yesiyu403 made their first contribution in #565
• @regend made their first contribution in #569
• @gyhbbb made their first contribution in #596
• @kaijchen made their first contribution in #601
• @shanemcd made their first contribution in #625
• @phagenlocher made their first contribution in #675
• @SKDG042 made their first contribution in #442
• @xuwuruoshui made their first contribution in #630
• @fanlv made their first contribution in #842
• @sheep44044 made their first contribution in #845
• @kklinan made their first contribution in #281
• @wucm667 made their first contribution in #793
• @Jmmt-mingrui made their first contribution in #831
• @kexyseeds made their first contribution in #860
• @shaxiaozz made their first contribution in #722
• @zbysir made their first contribution in #752
• @mjlzz made their first contribution in #872
• @marwan38 made their first contribution in #709
• @shaikustin made their first contribution in #888
• @sin-bufan made their first contribution in #929
• @Abtiotm made their first contribution in #938

Full Changelog: components/tool/mcp/v0.0.8...components/tool/mcp/v0.0.9