v2.2.59

What's Changed

• feat: Add Gemini internal tools examples by @MervinPraison in #917

Full Changelog: v2.2.58...v2.2.59

v2.2.58

What's Changed

• fix: add string LLM support to LLMGuardrail for Gemini models by @MervinPraison in #908
• fix: resolve litellm.BadRequestError with unrecognized output_json/output_pydantic parameters by @MervinPraison in #910
• Fix issue #904: Loop task callback inheritance and enhanced metadata support by @MervinPraison in #912
• fix: Add callback execution to synchronous display functions by @MervinPraison in #913
• Issue #901: Fix self-reflection with tools by @MervinPraison in #914
• fix: add execute and aexecute methods to Agent class by @MervinPraison in #915
• Fix: resolve duplicate callback execution in LLM interaction handling by @MervinPraison in #916

Full Changelog: v2.2.57...v2.2.58

v2.2.57

What's Changed

• fix: resolve Ollama sequential tool calling redundant execution issue by @MervinPraison in #895
• Issue #901: Changes from Claude by @MervinPraison in #902
• fix: exclude response_format parameter for Gemini image models by @MervinPraison in #903
• feat: enhance callbacks with agent and task context (fixes #896) by @MervinPraison in #897

Full Changelog: v2.2.56...v2.2.57

v2.2.56

What's Changed

• fix: prevent infinite loop in parallel task execution by @MervinPraison in #884
• feat: integrate Trafilatura for high-quality URL content extraction by @MervinPraison in #881
• fix: enable true parallel execution for async tasks in sequential mode by @MervinPraison in #883
• feat: add support for Gemini embedding models by @MervinPraison in #891
• fix: ensure callbacks work regardless of verbose setting by @MervinPraison in #886
• Add Natural_Language_Codebase notebook by @Dhivya-Bharathy in #898
• fix: handle single function as tools parameter to prevent iteration error by @MervinPraison in #894
• feat: Add Notion_Workspace_Assistant notebook by @Dhivya-Bharathy in #900

Full Changelog: v2.2.55...v2.2.56

v2.2.55

What's Changed

• fix: Implement Ollama sequential tool calling fix by @MervinPraison in #879
• feat: Add 10 Grok4 Intelligent Agent Examples by @Dhivya-Bharathy in #890
• feat: Add Gemini structured output support by @MervinPraison in #888
• fix: Support GuardrailResult return type in guardrail functions (#875) by @MervinPraison in #885
• fix: prevent duplicate callback invocations in LLM response generation by @MervinPraison in #887
• fix: use LLM abstraction instead of hardcoded OpenAI in hierarchical process by @MervinPraison in #889
• fix: Improve Ollama sequential tool calling handling by @MervinPraison in #892
• feat: Add multi-provider/multi-model support with intelligent routing by @MervinPraison in #880

Full Changelog: v2.2.54...v2.2.55

v2.2.54

What's Changed

• Fix #854: Ollama sequential tool calling failing by @praison-ai-bot[bot] in #856
• Fix #854: Ollama sequential tool calling failing by @praison-ai-bot[bot] in #857
• Fix #860: Hierarchical Process LLM Fallback by @praison-ai-bot[bot] in #861
• Fix #864: Fix parallel task loop by @praison-ai-bot[bot] in #865
• Fix #854: Ollama sequential still failing by @praison-ai-bot[bot] in #858
• Fix #862: Agent gets stuck in infinite loop when running multiple async tasks by @praison-ai-bot[bot] in #863

New Contributors

• @praison-ai-bot[bot] made their first contribution in #856

Full Changelog: v2.2.53...v2.2.54

v2.2.53

What's Changed

• fix: enable sequential tool execution in OpenAI client by @MervinPraison in #849
• fix: handle None response_text in Ollama sequential tool calls by @MervinPraison in #851

Full Changelog: v2.2.52...v2.2.53

v2.2.52

What's Changed

• PR #832: Changes from Claude by @MervinPraison in #834
• fix: capture final response after tool execution in LLM by @MervinPraison in #836
• fix: add return_dict=True to fix TypeError in prompt_chaining tests by @MervinPraison in #843
• docs: add OLLAMA_HANDLING.md to explain ollama_handled flag usage by @MervinPraison in #844
• fix: allow agents to continue with sequential tool calls by @MervinPraison in #842

Full Changelog: v2.2.51...v2.2.52

v2.2.51

What's Changed

• feat: Implement MCP HTTP Stream transport by @MervinPraison in #826
• fix: add validation feedback to workflow retry logic by @MervinPraison in #829
• fix: centralize logging configuration to respect LOGLEVEL environment variable by @MervinPraison in #833
• fix: sequential tool calling for non-streaming responses by @MervinPraison in #832

Full Changelog: v2.2.50...v2.2.51

v2.2.50

What's Changed

• fix: add missing HostedMCPServer class for MCP server implementations by @MervinPraison in #816
• Claude/issue 795 20250710 222918 by @MervinPraison in #812
• fix: enable tool calling for Gemini models by @MervinPraison in #819
• Add 10 New Grok4 Intelligent Agents Covering Diverse Domains by @Dhivya-Bharathy in #814

Full Changelog: v2.2.49...v2.2.50