v0.3.1

What's Changed

🛠 Enhancements made

• feat: propagate api_base property for local LLM support by @ahmednabiled in #239

🐛 Bugs fixed

• fix: apply_plan() flattens multiple tool call results by @khushiiagrawal in #155
• fix: prevent ReWOO replay from mutating cached tool_calls by @Harsh-617 in #160
• docs: fix incorrect api_key instruction in negotiation readme by @Justaanand50 in #249
• fix: guard against non-LLM recipients in speak_to tool(#158) by @BhoomiAgrawal12 in #161
• fix: consolidate evicted entries into long-term memory by @yashhzd in #166
• fix: render readable text in get_communication_history for all memory… by @Harsh-617 in #194
• fix: accumulate additive events in memory staging area by @yashhzd in #165
• fix: selected_tools=[] should return no tools, not all tools by @AdityaChauhanX07 in #225
• Fix: Raises value error for invalid tool name instead of key error by @DineshThumma9 in #280
• fix: change executor tool_choice from required to auto so reasoning respects no-action decisions by @saakshisinghal14 in #286

📜 Documentation improvements

• docs: improve README structure and fix duplicated section by @Indrasish7 in #235

🔧 Maintenance

• fix: replace deprecated seed= with rng= in tests by @R1patil in #226
• improve RateLimitError message with actionable guidance by @ospreyboi in #258
• remove unused is_speaking attribute from LLMAgent by @EloZiko in #261
• re-enable RUF012, B017, PGH004 linting rules and fix violations by @Komatlakarthik in #247
• pin mesa below 4 by @wang-boyu in #272

New Contributors

• @R1patil made their first contribution in #226
• @ospreyboi made their first contribution in #258
• @Indrasish7 made their first contribution in #235
• @Justaanand50 made their first contribution in #249
• @ahmednabiled made their first contribution in #239
• @EloZiko made their first contribution in #261
• @Komatlakarthik made their first contribution in #247
• @AdityaChauhanX07 made their first contribution in #225
• @DineshThumma9 made their first contribution in #280
• @saakshisinghal14 made their first contribution in #286

Full Changelog: v0.3.0...v0.3.1

v0.3.0

What's Changed

⚠️ Breaking changes

• fix: allow None for prompt in get_messages, generate, agenerate by @gamal1osama in #93

🛠 Enhancements made

• fix: store unique_ids instead of Agent objects in send_message memory… by @Harsh-617 in #157

🐛 Bugs fixed

• Fix async memory consolidation by @psbuilds in #69
• fix: async generation endpoint parity in ModuleLLM by @zhaosina in #105
• Fix critical vision radius bugs in LLMAgent perception system by @hillhack in #61
• fix: suppress spurious Auto-save failed warning from atexit in tests by @gamal1osama in #84
• fix: extract text content from LLM response in memory consolidation by @khansalman12 in #101
• Add ShortTermMemory Tests + BUG FIXES by @psbuilds in #76
• fix: align STLTMemory.get_prompt_ready() return type with Memory ABC by @khushiiagrawal in #118
• fix: CoT and ReWOO reasoning pass str to add_to_memory() where dict is expected by @khushiiagrawal in #123
• fix: remove duplicate space arg in ContinuousSpace teleport logic(#119) by @BhoomiAgrawal12 in #126
• fix: add missing obs and ttl params to ReWOOReasoning.plan() and apla… by @BhoomiAgrawal12 in #131
• fix: use agenerate_obs in ReWOOReasoning.aplan() instead of blocking generate_obs by @khansalman12 in #133
• Respect ignore_agent flag in docstring validation by @Harsh-617 in #130
• fix: prevent move_one_step from crashing with OrthogonalMooreGrid by @khushiiagrawal in #138
• Fix/Add Boundary Check in move_one_step to Prevent Cryptic Grid Errors by @gamal1osama in #146
• fix: store graded EpisodicMemory entries as MemoryEntry objects and use correct llm instance by @psbuilds in #109
• fix: handle non-LLM neighbors in _build_observation by @BhoomiAgrawal12 in #145

📜 Documentation improvements

• Fix broken tutorial link. by @divilian in #80
• docs: add orphaned pages to toctree in index.md by @yashhzd in #81
• reorganize readthedocs folder structure by @wang-boyu in #92
• update issue/pr templates and link to mesa contributor guide by @wang-boyu in #124

🔧 Maintenance

• Remove duplicate entries in all in mesa_llm.init by @gamal1osama in #94
• fix: replace deprecated AgentSet.getitem in tests by @yashhzd in #82
• remove redundant linter from GA and update ruff hook in pre-commit by @wang-boyu in #103
• fix(parallel_stepping): remove leftover debug print from _agentset_do… by @BhoomiAgrawal12 in #100
• refactor: replace rich console with standard logging and improve api … by @uday-codes69 in #85
• Validate llm_model format in ModuleLLM by @Harsh-617 in #89
• fix: replace deprecated AgentSet indexing for Mesa 4.0 compatibility by @Zain-Naqi in #110
• test: add memory × reasoning integration tests + fix async signature bugs by @yashhzd in #128
• test: consolidate and reuse test fixtures in conftest files by @wang-boyu in #121

New Contributors

• @psbuilds made their first contribution in #69
• @gamal1osama made their first contribution in #94
• @divilian made their first contribution in #80
• @yashhzd made their first contribution in #81
• @zhaosina made their first contribution in #105
• @hillhack made their first contribution in #61
• @BhoomiAgrawal12 made their first contribution in #100
• @uday-codes69 made their first contribution in #85
• @Harsh-617 made their first contribution in #89
• @khansalman12 made their first contribution in #101
• @Zain-Naqi made their first contribution in #110
• @khushiiagrawal made their first contribution in #118

Full Changelog: v0.2.0...v0.3.0

v0.2.0

What's Changed

🛠 Enhancements made

• Memory parent class improvements by @colinfrisch in #30

🐛 Bugs fixed

• Fix Ollama tool calling crash when system prompt is None by @februarysea in #41
• fix: treat ollama_chat as ollama provider by @februarysea in #43
• Fix prompt gestion by @colinfrisch in #48
• add error handling to move one step tool by @apoorvasj in #62

🔍 Examples updated

• Sugarscape g1mt by @sujalgawas in #35
• feat: Add dialogue history context to negotiation agents by @Nithin9585 in #56

📜 Documentation improvements

• update links for repository transfer by @wang-boyu in #19
• docs: Add Google Gemini to supported LLMs in README by @sujalgawas in #24
• Update organization name from projectmesa to mesa by @EwoutH in #39
• Update contact email. by @jackiekazil in #40
• docs: Add examples page and expand API documentation structure by @colinfrisch in #33
• Add mesa-llm overview by @niteshver in #60
• Add first mesa-llm model tutorial by @niteshver in #45
• add getting_started for mesa_llm by @niteshver in #64
• Implemented API documentation using Sphinx by @pushpitkamboj in #46
• docs: add sugarscrap model in example folder by @niteshver in #74

🔧 Maintenance

• test(reasoning): Add tests for Reasoning base class by @DipayanDasgupta in #21
• Test episodic memory by @sujalgawas in #26
• Test lt memory by @sujalgawas in #27
• test: update module_llm tests for system prompt handling (None -> "") by @februarysea in #44
• configure automatically generated release notes by @wang-boyu in #72

Other changes

• Fix and add dependencies by @colinfrisch in #49

New Contributors

• @DipayanDasgupta made their first contribution in #21
• @sujalgawas made their first contribution in #24
• @EwoutH made their first contribution in #39
• @jackiekazil made their first contribution in #40
• @februarysea made their first contribution in #41
• @Nithin9585 made their first contribution in #56
• @niteshver made their first contribution in #60
• @apoorvasj made their first contribution in #62
• @pushpitkamboj made their first contribution in #46

Full Changelog: v0.1.1...v0.2.0