Skip to content

Release history

agentscope releases

Build and run agents you can see, understand and trust.

Back to tool Latest release

All releases

14 shown

Review required
v2.0.0 Breaking risk
Auth RBAC Breaking upgrade

msg, tool, mcp, model, middleware

Open
No immediate action
v1.0.21 Mixed

Reasoning effort + Gemini signature + Formatter fix

Open
No immediate action
v1.0.20 Security relevant

Routine maintenance and dependency updates.

patches CVE-2026-42208
Open
v1.0.19.post1 New feature
Notable features
  • Execution timeout for MCP operations
  • Error handling for OpenAI streaming responses
Full changelog

What's Changed

  • fix(openai): catch response_format errors during streaming to enable fallback by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1527
  • feat(mcp): add execution timeout for mcp by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1540
  • chore(version): bumping version to 1.0.19.post1 by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1543

Full Changelog: https://github.com/agentscope-ai/agentscope/compare/v1.0.19...v1.0.19.post1

View release on GitHub
v1.0.19 Mixed
Notable features
  • Tablestore memory support for alternative persistence backend
  • Local file path support (file://) in formatters with automatic base64 conversion
  • Structured output fallback for OpenAI chat model
Full changelog

What's Changed

  • docs(copaw): add news for copaw by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1379
  • feat(formatter): support local multimedia paths starting with file:// by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1385
  • feat(memory): add tablestore memory support by @zhanghaochenFCB in https://github.com/agentscope-ai/agentscope/pull/1308
  • fix(examples): rename ReflectFailure.decompose_subtask.rephrased_plan to failed_subtask in deep research example by @alex-xinlu in https://github.com/agentscope-ai/agentscope/pull/1151
  • docs(news): update the news for 2.0 by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/1442
  • chore(assets): update the png by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/1468
  • fix(agent): add missing await on memory.add in DeepResearchAgent._summarizing by @Ricardo-M-L in https://github.com/agentscope-ai/agentscope/pull/1477
  • fix: handle missing 'type' field in tool parameter properties for OpenAITokenCounter by @octo-patch in https://github.com/agentscope-ai/agentscope/pull/1403
  • fix(agent): handle thinking blocks in DeepResearchAgent to support thinking and non-thinking models by @Luohh5 in https://github.com/agentscope-ai/agentscope/pull/1492
  • fix(memory): serialize concurrent writes in AsyncSQLAlchemyMemory by @umerkhan95 in https://github.com/agentscope-ai/agentscope/pull/1390
  • fix(agent): prevent duplicate hook execution when subclass overrides _reasoning or _acting by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1481
  • feat(model): add fallback for structured output for openai chat model by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1430
  • fix(sqlalchemy_memory): flush the session to ensure message records are written to the database by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1396
  • feat(formatter): translate local file path to base64 in image blocks in Anthropic formatter by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1361
  • fix: use underscore naming for Tavily tool functions in DeepResearch agent by @octo-patch in https://github.com/agentscope-ai/agentscope/pull/1400
  • feat(model): refine counting input tokens for anthropic chat model by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1319
  • fix(DashScope formatter): convert local files to base64 data in dashscope formatter by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1253
  • fix(agent): guard DeepResearchAgent tool metadata access by @YingchaoX in https://github.com/agentscope-ai/agentscope/pull/1489
  • fix(unittest): hotfix for unittest by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1507
  • chore(version): bumping verstion to 1.0.19 by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1506

New Contributors

  • @zhanghaochenFCB made their first contribution in https://github.com/agentscope-ai/agentscope/pull/1308
  • @alex-xinlu made their first contribution in https://github.com/agentscope-ai/agentscope/pull/1151
  • @Ricardo-M-L made their first contribution in https://github.com/agentscope-ai/agentscope/pull/1477
  • @octo-patch made their first contribution in https://github.com/agentscope-ai/agentscope/pull/1403
  • @umerkhan95 made their first contribution in https://github.com/agentscope-ai/agentscope/pull/1390

Full Changelog: https://github.com/agentscope-ai/agentscope/compare/v1.0.18...v1.0.19

View release on GitHub
v1.0.18 Mixed
Notable features
  • Tuner enhanced with prompt tuning and model selection
  • Toolkit supports background execution for tools
  • Formatter improvements: local image URLs without extension and file:// audio paths
Full changelog

What's Changed

  • fix(huggingface): add huggingface endpoint fix by @garyzhang99 in https://github.com/agentscope-ai/agentscope/pull/1340
  • docs(readme): update the README.md and README_zh.md with copaw by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/1353
  • feat(test): fix version of pymilvus by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1355
  • feat(formatter): enhance _to_openai_image_url for local files without extension by @kingofpowers in https://github.com/agentscope-ai/agentscope/pull/1341
  • feat(tuner): enhance tuner with prompt tuning and model selection by @Kanaricc in https://github.com/agentscope-ai/agentscope/pull/1190
  • fix(tuner): set monitor_type on monitor config instead of overwriting monitor object by @PatrickG1014 in https://github.com/agentscope-ai/agentscope/pull/1365
  • feat(toolkit): support background execution for tools by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/1342
  • feat(agent): filter thinking content before broadcast by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1332
  • feat(formatter): supprot audio path start with file:// in OpenAI formatter by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1349
  • docs: update github trending image by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1368
  • fix(model): avoid streaming tool input regression by @YingchaoX in https://github.com/agentscope-ai/agentscope/pull/1356
  • feat(model): pass through the model's response id by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1333
  • chore(version): update version to 1.0.18 for release by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1370

New Contributors

  • @kingofpowers made their first contribution in https://github.com/agentscope-ai/agentscope/pull/1341
  • @Kanaricc made their first contribution in https://github.com/agentscope-ai/agentscope/pull/1190

Full Changelog: https://github.com/agentscope-ai/agentscope/compare/v1.0.17...v1/0.18

View release on GitHub
v1.0.16 Bug fix
Notable features
  • DashScope CosyVoice TTS model support
  • Toolkit middleware for tool execution control
  • OpenAI Realtime model tool format support
View release on GitHub
v1.0.15 New feature
Notable features
  • Realtime voice agent implementation
  • OceanBaseStore as vector database option for RAG
  • Logger and Tinker support in agent tuner
Full changelog

What's Changed

  • feat(tuner): Add Logger and Tinker Support by @pan-x-c in https://github.com/agentscope-ai/agentscope/pull/1178
  • feat(rag/store): add OceanBaseStore as a vdb choice by @PsiACE in https://github.com/agentscope-ai/agentscope/pull/1078
  • feat(voice agent): implement realtime voice agent by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1056
  • docs(evaluation): add openjudge evaluation integration by @ployts in https://github.com/agentscope-ai/agentscope/pull/1168
  • docs(api): abandon the API documentation temporarily by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/1196
  • docs(tutorial): remove the huggingface related code in tutorial to avoid rate limit by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/1198
  • fix(tutorial): fix the bug in the tutorial by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/1199
  • fix(mcp): fix the timeout argument in MCPToolFunction class by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/1192
  • Adapt Mem0LongtermMemory to mem0 v1 by @Osier-Yi in https://github.com/agentscope-ai/agentscope/pull/1191
  • fix(gemini): fix GeminiChatModel compatibility with Gemini-3-pro by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/1206
  • chore(version): update version to 1.0.15 for new release by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/1209
  • fix(redis_memory): support decode_response to be False in the given connection_pool argument by @ceshihao in https://github.com/agentscope-ai/agentscope/pull/1187
  • perf(redis_memory): avoid scan operations in RedisMemory to improve efficiency by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/1211

New Contributors

  • @PsiACE made their first contribution in https://github.com/agentscope-ai/agentscope/pull/1078
  • @ployts made their first contribution in https://github.com/agentscope-ai/agentscope/pull/1168

Full Changelog: https://github.com/agentscope-ai/agentscope/compare/v1.0.14...v1.0.15

View release on GitHub
v1.0.14 New feature
Notable features
  • Disable tool use input parsing via stream_tool_parsing flag
  • Redis-based session support
  • PowerPoint reader for RAG systems
View release on GitHub
v1.0.13 Mixed
Notable features
  • MongoDB vector store support in RAG module
  • Excel reader for document processing in RAG module
  • Qwen-Deep-Research model integration example
Full changelog

Highlight

RAG

  • Support MongoDB vector database
  • Support Excel reader

Example

  • Integration with Qwen-Deep-Research model

What's Changed

  • Add example integration for Qwen-Deep-Research model by @denverdino in https://github.com/agentscope-ai/agentscope/pull/869
  • feat(tool): add func_name parameter to support custom tool renaming by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1115
  • feat(rag): add MongoDB vector store in the RAG module. by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/860
  • ci(gemini): add code review guide for the gemini code assist by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/1135
  • fix(doc): fix typos by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1137
  • feat(rag): add excel reader to support excel files in the rag module by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/872
  • docs: update the documentations in the AgentScope repo by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/1128
  • fix(formattter): assigns content as an empty list [] for messages without valid content blocks by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1141
  • fix(model): fix openai stream parse when delta is None by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1142
  • chore(version): update the version to 1.0.13 for a quick release by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/1143
  • feat(dashscope): support to load headers from the environment by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/1096

New Contributors

  • @denverdino made their first contribution in https://github.com/agentscope-ai/agentscope/pull/869

Full Changelog: https://github.com/agentscope-ai/agentscope/compare/v1.0.12...v1.0.13

View release on GitHub
v1.0.12 New feature
Notable features
  • ReActAgent automatic memory compression
  • Database memory with relational DB and Redis support
  • Azure OpenAI API compatibility
View release on GitHub
v1.0.11 Mixed
Notable features
  • A2A protocol support for inter-agent communication
  • Memory recording before agent reply generation
Full changelog

What's Changed

  • add record_to_memory before generate return reply_message by @Osier-Yi in https://github.com/agentscope-ai/agentscope/pull/1039
  • fix(werewolf game): Broadcast game result when game ends at night phase by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1065
  • fix(openai formatter): Correctly extract image extension from URLs with query parameters by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1066
  • feat(agent): Add A2A protocol support for inter-agent communication by @Sunrisea in https://github.com/agentscope-ai/agentscope/pull/1027
  • raise exception on MCP disconnection by @MengsD in https://github.com/agentscope-ai/agentscope/pull/1024
  • fix(gemini model): flatten json schema in tools for gemini chat model by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/1050
  • bugfix: parse anthropic response content block type logic error. by @FFengIll in https://github.com/agentscope-ai/agentscope/pull/1070
  • chore(version): update version to 1.0.11 by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/1074

New Contributors

  • @Sunrisea made their first contribution in https://github.com/agentscope-ai/agentscope/pull/1027
  • @MengsD made their first contribution in https://github.com/agentscope-ai/agentscope/pull/1024
  • @FFengIll made their first contribution in https://github.com/agentscope-ai/agentscope/pull/1070

Full Changelog: https://github.com/agentscope-ai/agentscope/compare/v1.0.10...v1.0.11

View release on GitHub

Beta — feedback welcome: [email protected]