AgentScope Files

What's Changed

• fix(gemini): remove unsupported $schema from tool parameters by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/2277
• feat(workspace): share workspace across different agents by @iluv7 in https://github.com/agentscope-ai/agentscope/pull/1951
• feat(workspace): isolate skills per agent, select them per session by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2283
• fix(hub): preserve GitHub MCP environment inputs by @RerankerGuo in https://github.com/agentscope-ai/agentscope/pull/2230
• fix(app): stop stranding session inbox payloads until the next user turn by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2295
• feat(console): add terminal console for trying and debugging agents by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2297
• fix(tool): fix the meta tool by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2302
• fix(model): fixing MoonshotChatModel streaming usage by @silicon14 in https://github.com/agentscope-ai/agentscope/pull/2314
• fix(mcp): allow reconnecting stateful clients by @Solaris-star in https://github.com/agentscope-ai/agentscope/pull/2308
• feat(agent): let on_reply middlewares swallow ReplyEndEvent to continue the reply loop by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2322
• fix(formatter): use data URLs for DashScope base64 audio by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/2315
• fix(model): close OpenAI Responses stream deterministically by @BlessedWithLuck1105 in https://github.com/agentscope-ai/agentscope/pull/2349
• fix(model): close Anthropic stream deterministically by @BlessedWithLuck1105 in https://github.com/agentscope-ai/agentscope/pull/2350
• fix(formatter): keep images accumulated before a tool call by @BlessedWithLuck1105 in https://github.com/agentscope-ai/agentscope/pull/2351
• fix(hub): accept owner-scoped skill card ids by @csurong in https://github.com/agentscope-ai/agentscope/pull/2291
• feat(rag): make chunker config explicit and schema-driven by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/2083
• fix(ui): distinguish credentials by name in TTS and fallback model selectors by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/2169
• fix: disable thinking for structured output calls and add compression fallback by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/2140
• fix(app): load mutable session state under session lock by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/2232
• feat(agent): add max_image_num to ContextConfig to limit images in context by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2362
• fix(tool): pick the shell by the backend OS instead of the host OS in Bash by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2366
• docs(tool): fix typo in toolkit group validation comment by @hata333 in https://github.com/agentscope-ai/agentscope/pull/2355
• feat(tool): support DataBlock return in Read tool for binary and structured files by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/2114
• fix(tool): resolve PEP 563 lazy annotations when extracting tool schemas by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2371
• fix(event): validate tool result data source by @fancyboi999 in https://github.com/agentscope-ai/agentscope/pull/2370
• fix(mcp): detect SSE transport by URL path so query strings work by @1753936488 in https://github.com/agentscope-ai/agentscope/pull/2311
• fix(tool): distinguish inactive tool groups from missing tools in check_tool_available by @wizardeur in https://github.com/agentscope-ai/agentscope/pull/2316
• fix(app): clean up index worker tasks on cancellation by @helloJamest in https://github.com/agentscope-ai/agentscope/pull/2220
• fix(utils): make DictMixin honor missing-attribute semantics by @jesseedcp in https://github.com/agentscope-ai/agentscope/pull/2241
• fix(state): refresh read file cache recency on access by @faisalishfaq2005 in https://github.com/agentscope-ai/agentscope/pull/1811
• fix(memory): own minimal ReMe configuration by @ployts in https://github.com/agentscope-ai/agentscope/pull/2374
• feat(tool): support custom input_schema in FunctionTool by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2378
• feat(ui): display exceeding max iteration alert in chat page by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2381
• feat(rag): add chunk browsing, raw document preview and enriched knowledge base listing by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2372
• feat(agent-team): add loop middleware for subagent in teams by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2379
• feat(agent-team): tell the leader when a member's turn fails by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2386
• fix(web_ui): keep an uploaded file visible while the worker indexes it by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2387
• fix(rag): escape Word and PPT Markdown table cells by @kaiwangleo in https://github.com/agentscope-ai/agentscope/pull/2377
• fix(agent): keep prompt cache tokens in usage by @yingz23 in https://github.com/agentscope-ai/agentscope/pull/2318
• feat(channel): hold the long connections in a dedicated worker by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2390
• fix(context): allow trigger ratio of 0.9 by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/2396
• refactor(channel): deliver a reply from the node that produced it by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2395
• feat(channel): add DingTalk channel by @iluv7 in https://github.com/agentscope-ai/agentscope/pull/2285
• chore(version): bump version to 2.0.7 for release by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2407
• feat(workspace): add workspace pool for pooling by @cuiyuebing in https://github.com/agentscope-ai/agentscope/pull/1755
• fix(tracing): include input messages in chat spans by @qbc2016 in https://github.com/agentscope-ai/agentscope/pull/2391
• fix(scheduler): bind a real workspace to scheduled sessions by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2410
• fix(channel): let DingTalk send without the inbound connection by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2409
• feat(storage): support channel records on SQL storage by @DavdGao in https://github.com/agentscope-ai/agentscope/pull/2388

New Contributors

• @silicon14 made their first contribution in https://github.com/agentscope-ai/agentscope/pull/2314
• @Solaris-star made their first contribution in https://github.com/agentscope-ai/agentscope/pull/2308
• @BlessedWithLuck1105 made their first contribution in https://github.com/agentscope-ai/agentscope/pull/2349
• @hata333 made their first contribution in https://github.com/agentscope-ai/agentscope/pull/2355
• @1753936488 made their first contribution in https://github.com/agentscope-ai/agentscope/pull/2311
• @wizardeur made their first contribution in https://github.com/agentscope-ai/agentscope/pull/2316
• @jesseedcp made their first contribution in https://github.com/agentscope-ai/agentscope/pull/2241
• @kaiwangleo made their first contribution in https://github.com/agentscope-ai/agentscope/pull/2377
• @yingz23 made their first contribution in https://github.com/agentscope-ai/agentscope/pull/2318

Full Changelog: https://github.com/agentscope-ai/agentscope/compare/v2.0.6...v2.0.7