What's Changed

• Update question.yml by @sscargal in #890
• Always pull the latest MemMachine Docker Image by @sscargal in #889
• Fixes #886 - memmachine-compose.sh ignores MEMMACHINE_IMAGE in .env by @sscargal in #887
• Docs: Updating REST Example to correctly represent V2 API Endpoints by @SarahScargall in #876
• Implement score threshold for search by @edwinyyyu in #898
• Fix: Support local images by relaxing pull policy when remote pull fails by @sscargal in #900
• Add new shields to track downloads by @sscargal in #905
• feat: support configurable MEMMACHINE_WORKERS and fix circular import by @sscargal in #903
• Issue#841 support restapi in locomo benchmark by @tomw-mv in #843
• Move function _list_target_memories to service.py by @xiongzubiao in #891
• Bump ruff from 0.14.7 to 0.14.10 by @dependabot[bot] in #896
• Bump langchain-aws from 1.1.0 to 1.2.0 by @dependabot[bot] in #895
• Bump tzdata from 2025.2 to 2025.3 by @dependabot[bot] in #893
• feat: Add MemMachine Integration Guide for crewai. by @tianAndrew in #875
• Change dify plugin label by @mwqgithub in #907
• Migrate from pickle to JSON by @malatewang in #927
• Improve robustness and fix timing issues by @sscargal in #932
• Fix postgres migration problem by @malatewang in #933
• feat: Keep all components aligned on the same version at any commit. by @tianAndrew in #866
• Add TypeScript REST client by @yaleMemVerge in #339
• Docs: Adding Dify Integration Guide by @SarahScargall in #881
• Remove duplicated mcp mount by @xiongzubiao in #925
• Fix misc documents by @honggyukim in #919
• Update Contributing Guides by @sscargal in #924
• Ensure only the CPU image is tagged with 'latest' by @sscargal in #935
• Bump aiosqlite from 0.22.0 to 0.22.1 by @dependabot[bot] in #894
• Fix CodeQL 'Information exposure through an exception' issues by @sscargal in #928
• Docs: New TypeScript SDK for our OpenSource API by @SarahScargall in #838
• Fix cartesian product performance notifications caused by the comma-separated MATCH pattern in Cypher queries by @sscargal in #914
• Add FastGPT Integration by @yaleMemVerge in #934
• Qwen agent integration by @mwqgithub in #921

New Contributors

• @yaleMemVerge made their first contribution in #339

Full Changelog: v0.2.2...v0.2.3

What's Changed

• Docs: Resolving Issue #766 - rename of .mcp.json to claude_desktop_config.json by @SarahScargall in #790
• Add coding style prompt by @o-love in #756
• fix: update API endpoint URLs in startup script by @dxu104 in #803
• feat: aws strands integration. by @tianAndrew in #528
• Fix episodic configuration is not set correctly by @malatewang in #791
• Unifying docker-compose and docker compose by @honggyukim in #774
• Fix URLs when generating openapi.json by @sscargal in #804
• Add test pkg ut by @tianAndrew in #752
• [GH-703] Semantic: Add ingestion trigger configuration. by @o-love in #755
• fix: update MemMachine API URL to /docs endpoint by @dxu104 in #811
• Fix main memmachine semantic ingestion test by @o-love in #816
• Docs: New Feature - Platform Section by @SarahScargall in #813
• Change memory type in docs to lowercase by @mwqgithub in #807
• Add episodic_type to add memory api by @mwqgithub in #796
• Remove trace from known error by @jealous in #794
• Examples cleanup by @gauri-nagavkar-memverge in #821
• Add AGENTS.MD by @Charlie-Yi-2002 in #814
• Add info logs to semantic memory ingestion by @o-love in #824
• feat: update python client. by @tianAndrew in #828
• Fix Type issues in main branch by @o-love in #832
• allow supply credentials with env reference by @jealous in #819
• Fix the summary prompt not initialized properly by @malatewang in #812
• Add metrics to AmazonBedrockReranker by @edwinyyyu in #798
• Run unit tests in parallel by @o-love in #831
• [GH-713] Support more timestamp format. by @jealous in #801
• Can not get semantic memory with MCP by @jealous in #815
• [GH-701] Validate embedder, reranker and language model by @jealous in #802
• [GH-712] Fix race condition in resource manager by @jealous in #839
• Fix YAML validation issue in Question template by @sscargal in #842
• Fix episode deletion in short term memory by @malatewang in #833
• Fix neo4j initialization problem by @malatewang in #834
• Add openai-compatible model provider in memmachine-compose.sh by @mwqgithub in #827
• Resolving link issues in docs.json by @SarahScargall in #858
• Potential fix for code scanning alert no. 37: Workflow does not contain permissions by @sscargal in #777
• Return an empty response for empty episodic memories by @mwqgithub in #859
• Wrap startup DB failures as ConfigurationError by @mwqgithub in #870
• Ensure docker build fails if NLTK download fails by @mwqgithub in #869
• List both episodic and semantic memory if type is not specified by @mwqgithub in #867
• Return 404 when deleting nonexistent projects by @mwqgithub in #868
• Add dify integration by @mwqgithub in #820
• fix: support timestamp when adding memory using python client sdk by @tianAndrew in #861
• Fix apache license copyright information by @o-love in #873
• Bump cohere from 5.20.0 to 5.20.1 by @dependabot[bot] in #852
• Bump sentence-transformers from 5.1.2 to 5.2.0 by @dependabot[bot] in #853
• Bump openai from 2.8.1 to 2.14.0 by @dependabot[bot] in #854
• Bump aiosqlite from 0.21.0 to 0.22.0 by @dependabot[bot] in #855
• Fix URL in platform API docs by @xiongzubiao in #880
• Implement embedder and reranker input length limits and sanitization by @edwinyyyu in #744
• [GH-778] Fix response doesn't have metric dictionary by @o-love in #885
• Disable message sentence chunking by default by @edwinyyyu in #874

New Contributors

• @honggyukim made their first contribution in #774

Full Changelog: v0.2.1...v0.2.2

What's Changed

• Docs: Adding "memmachine-server" and "memmachine-client" packages where appropriate in pip install commands. by @SarahScargall in #736
• Rename root project to 'memmachine-dev' to avoid namespace conflict with the 'memmachine' package by @sscargal in #745
• Revert "Rename root project to 'memmachine-dev' to avoid namespace co… by @sscargal in #747
• missing session prompt in default configuration by @jealous in #760
• Add Question Issue Template by @sscargal in #748
• Fix memory type examples by @xiongzubiao in #762
• Docs: Updating our Install Guide and Quick Start Guide to reflect two packages and more refined download instructions by @SarahScargall in #757
• Add support for filter comparisons by @o-love in #676
• Feat: Cohere reranker by @vinares in #761
• Ensure the CPU Docker Image is tagged with 'latest' by @sscargal in #776
• feat: add llamaindex integrations by @Lovegoodstudy in #746
• Docs: Openapi.json: updating the server url for correct port by @SarahScargall in #769
• Bump openai from 2.6.1 to 2.8.1 by @dependabot[bot] in #583
• [GH-706] Validate log level is applied by @o-love in #779
• Simple chatbot using v2 API endpoints by @gauri-nagavkar-memverge in #732
• Only ask for base url once in memmachine-compose.sh by @mwqgithub in #751
• Potential fix for code scanning alert no. 38: Clear-text logging of sensitive information by @sscargal in #788
• Deduplicate derivatives' source episodes by @edwinyyyu in #710

New Contributors

• @vinares made their first contribution in #761
• @Lovegoodstudy made their first contribution in #746

Full Changelog: v0.2.0...v0.2.1

What's Changed

• Claude code Integration document by @PeterZZZ in #530
• Implement script to dump Episodes for migrations from v0.1.x by @edwinyyyu in #586
• feat: build separate pkgs by @tianAndrew in #527
• Merge the development branch to main by @malatewang in #631
• Docsv2 by @SarahScargall in #635
• Ask for api endpoint in memmachine-compose.sh by @mwqgithub in #638
• Add unit test for api v2 by @mwqgithub in #632
• Add gitstar by @ChristianKniep in #285
• fix memory typo link by @ChristianKniep in #642
• Bump ruff from 0.14.5 to 0.14.7 by @dependabot[bot] in #655
• Bump boto3 from 1.40.74 to 1.42.0 by @dependabot[bot] in #653
• Bump fastapi from 0.121.2 to 0.123.0 by @dependabot[bot] in #656
• generate v2 config file with user inputs by @jgong in #659
• Update asyncpg version. by @sscargal in #648
• Add episodic storage message count to top level MemMachine by @o-love in #652
• fix: can not import MemMachine client. by @tianAndrew in #639
• fix: rename history_id to episode_id in EpisodeStorage.get_episode by @dxu104 in #647
• fix: integration test requires api key by @tianAndrew in #670
• fix: fix ci concurrency issue. by @tianAndrew in #671
• skip: skip client-integration check from forked branch. by @tianAndrew in #672
• Return prometheus metrics factory by default by @jealous in #667
• Add count cache for EpisodeStorage by @o-love in #662
• Add get episode count api. by @mwqgithub in #668
• Update slow integration test for MemMachine by @o-love in #674
• Sanitize text for AmazonBedrockReranker by @edwinyyyu in #646
• Update pypi build process by @sscargal in #650
• Docs: Updating Resources for V2 by @SarahScargall in #677
• fix dockerfile by copying required files by @jgong in #686
• CI fix: Downgrade ty by @o-love in #687
• Docs: Resolving mermaid issue in flow chart by @SarahScargall in #688
• Docs: Updating Hello World and Output for our PythonSDK Client by @SarahScargall in #689
• Fix MCP search_memory filter to use metadata.user_id (BUG #696) by @dxu104 in #698
• build with gpu using memmachine-compose.sh build by @jgong in #699
• CI Fix: Disable bedrock llm test in PR and bedrock llm fixes by @o-love in #697
• Escape episode content when returning episode context as string by @edwinyyyu in #660
• add installation and configuration script. by @jealous in #682
• Fix MCP stdio output to keep stdout JSON-only # by @dxu104 in #707
• Update mcp document. by @jealous in #708
• docs: Add Docker Compose setup as recommended MCP configuration method by @dxu104 in #711
• Add document for REST API v2. by @jealous in #715
• Add document for configuration wizard. by @jealous in #718
• Explicitly run MemMachine on all IP addresses (0.0.0.0). by @sscargal in #723
• [GH-709] Neo4j Semantic Memory: Lock concurrent index creation by @o-love in #719
• Docs: Updating PythonSDK Server Hello World, New Install Guide Cloud Deployments Section, Linking to new Configure Wizard by @SarahScargall in #725
• Support delete multiple memories. by @jealous in #724
• Implement RestError to handle API error. by @jealous in #721
• Do not exit if user does not install neo4j by @jealous in #720
• Merge the add memory apis into one. by @mwqgithub in #730
• Treat : as valid character in id by @jealous in #731
• set model name correctly in configuration.yml by @jgong in #733
• feat: update python sdk add memory method. by @tianAndrew in #734
• Update openapi.json by @sscargal in #735
• [GH-678] Semantic: Ignore invalid column names in filter by @o-love in #722
• Create a 'memmachine' metapackage for 'memmachine-client' and 'memmachine-server' by @sscargal in #737
• Bugfix/uv lock by @sscargal in #738
• Use the latest version of uv by @sscargal in #740
• Chore/build docker by @sscargal in #742
• Replace action module with manual cleanup of the root filesystem to free space by @sscargal in #743
• correct the env name for mcp param override by @jealous in #741

Full Changelog: v0.1.10...v0.2.0

What's Changed

• [GH-678] Semantic: Ignore invalid column names in filter by @o-love in #722
• Create a 'memmachine' metapackage for 'memmachine-client' and 'memmachine-server' by @sscargal in #737
• Bugfix/uv lock by @sscargal in #738
• Use the latest version of uv by @sscargal in #740

Full Changelog: v0.2.0-rc1...v0.2.0-rc2

What's Changed

• Claude code Integration document by @PeterZZZ in #530
• Implement script to dump Episodes for migrations from v0.1.x by @edwinyyyu in #586
• feat: build separate pkgs by @tianAndrew in #527
• Merge the development branch to main by @malatewang in #631
• Docsv2 by @SarahScargall in #635
• Ask for api endpoint in memmachine-compose.sh by @mwqgithub in #638
• Add unit test for api v2 by @mwqgithub in #632
• Add gitstar by @ChristianKniep in #285
• fix memory typo link by @ChristianKniep in #642
• Bump ruff from 0.14.5 to 0.14.7 by @dependabot[bot] in #655
• Bump boto3 from 1.40.74 to 1.42.0 by @dependabot[bot] in #653
• Bump fastapi from 0.121.2 to 0.123.0 by @dependabot[bot] in #656
• generate v2 config file with user inputs by @jgong in #659
• Update asyncpg version. by @sscargal in #648
• Add episodic storage message count to top level MemMachine by @o-love in #652
• fix: can not import MemMachine client. by @tianAndrew in #639
• fix: rename history_id to episode_id in EpisodeStorage.get_episode by @dxu104 in #647
• fix: integration test requires api key by @tianAndrew in #670
• fix: fix ci concurrency issue. by @tianAndrew in #671
• skip: skip client-integration check from forked branch. by @tianAndrew in #672
• Return prometheus metrics factory by default by @jealous in #667
• Add count cache for EpisodeStorage by @o-love in #662
• Add get episode count api. by @mwqgithub in #668
• Update slow integration test for MemMachine by @o-love in #674
• Sanitize text for AmazonBedrockReranker by @edwinyyyu in #646
• Update pypi build process by @sscargal in #650
• Docs: Updating Resources for V2 by @SarahScargall in #677
• fix dockerfile by copying required files by @jgong in #686
• CI fix: Downgrade ty by @o-love in #687
• Docs: Resolving mermaid issue in flow chart by @SarahScargall in #688
• Docs: Updating Hello World and Output for our PythonSDK Client by @SarahScargall in #689
• Fix MCP search_memory filter to use metadata.user_id (BUG #696) by @dxu104 in #698
• build with gpu using memmachine-compose.sh build by @jgong in #699
• CI Fix: Disable bedrock llm test in PR and bedrock llm fixes by @o-love in #697
• Escape episode content when returning episode context as string by @edwinyyyu in #660
• add installation and configuration script. by @jealous in #682
• Fix MCP stdio output to keep stdout JSON-only # by @dxu104 in #707
• Update mcp document. by @jealous in #708
• docs: Add Docker Compose setup as recommended MCP configuration method by @dxu104 in #711
• Add document for REST API v2. by @jealous in #715
• Add document for configuration wizard. by @jealous in #718
• Explicitly run MemMachine on all IP addresses (0.0.0.0). by @sscargal in #723
• [GH-709] Neo4j Semantic Memory: Lock concurrent index creation by @o-love in #719
• Docs: Updating PythonSDK Server Hello World, New Install Guide Cloud Deployments Section, Linking to new Configure Wizard by @SarahScargall in #725
• Support delete multiple memories. by @jealous in #724
• Implement RestError to handle API error. by @jealous in #721
• Do not exit if user does not install neo4j by @jealous in #720
• Merge the add memory apis into one. by @mwqgithub in #730
• Treat : as valid character in id by @jealous in #731
• set model name correctly in configuration.yml by @jgong in #733
• feat: update python sdk add memory method. by @tianAndrew in #734
• Update openapi.json by @sscargal in #735

Full Changelog: v0.1.10...v0.2.0-rc1

What's Changed

• Feat: support AWS session token by @Kevin-K-W in #420
• Feat: add client for ADE integration by @tianAndrew in #490
• Update embedders to use BaseModel params by @edwinyyyu in #415
• Fix OpenAIEmbedder used in integration test by @edwinyyyu in #494
• Bug: hackthon chatgpt2memmachine not converting time string to timestamp #497 by @tomw-mv in #498
• Fix cross_encoder typo by @edwinyyyu in #499
• fix values for aws_model and aws_embedder_id by @jgong in #500
• Simplify parameters for LongTermMemory, DeclarativeMemory by @edwinyyyu in #489
• Add deprecated check with ruff by @o-love in #510
• feat. support string format parameters by @PeterZZZ in #501
• Removed Node v18 test from "docs-checks" due to EOL status by @sscargal in #521
• doc. Fix errors in mcp document. by @jealous in #518
• generate configuration.yml with selected provider by @jgong in #517
• Docs: Update to openapi.json (API Reference Playground) for new MCP Tools from app.py by @SarahScargall in #486
• handles config for vector_graph_store by @jgong in #515
• Docs: Updating Memory Tables and Nomenclature from "Profile" to "Semantic" by @SarahScargall in #526
• Update language models to use BaseModel params by @edwinyyyu in #413
• feat: add langgraph support by @tianAndrew in #491
• Docs: Adding new PythonSDK Client Section and Page by @SarahScargall in #496
• fix ruff check errors. by @jealous in #531
• Fix deleting session episodes v1 by @edwinyyyu in #532
• Updating language to acknowledge multiple databases can be used by @SarahScargall in #529
• Add Playground link to navbar pointing to Hugging Face Spaces by @Anirudhem-mv in #561
• Docs: Add OpenAI GPT Store integration guide by @dxu104 in #564
• Bump fastmcp from 2.13.0.1 to 2.13.0.2 by @dependabot[bot] in #445

New Contributors

• @Kevin-K-W made their first contribution in #420
• @PeterZZZ made their first contribution in #501
• @Anirudhem-mv made their first contribution in #561

Full Changelog: v0.1.9...v0.1.10

What's Changed

• feat. Implement a simple MCP server by @jealous in #405
• Update Profile memory for Bedrock and Ollama by @gauri-nagavkar-memverge in #441

Full Changelog: v0.1.8...v0.1.9

What's Changed

• Quick fix for leaked resources by @edwinyyyu in #365
• Fix type check error by @malatewang in #366
• Add sentence transformer embedder support by @edwinyyyu in #367
• Embedder Config by @ChristianKniep in #357
• fixed 400 bad request error when calling chat completions by @scottly06 in #341
• consistent variable name for model in config and code, add ollama config in sample configs, add bedrock and ollama support in memmachine-compose.sh by @gauri-nagavkar-memverge in #372
• Add Documentation Checks and Validation Workflow by @sscargal in #347
• Fix build-docker.sh for macos compatibility by @jgong in #355
• Embedder model config by @gauri-nagavkar-memverge in #388
• Feat: add migration tool from hackathon by @tomw-mv in #231
• Update rerankers to use BaseModel params by @edwinyyyu in #368
• Update DeclarativeMemory resources to use BaseModel params by @edwinyyyu in #373
• Update Neo4jVectorGraphStore params BaseModel by @edwinyyyu in #350
• Disable sentence-transformers progress bars by @edwinyyyu in #376
• Bump ruff from 0.14.1 to 0.14.2 by @dependabot[bot] in #383
• Bump boto3 from 1.40.55 to 1.40.59 by @dependabot[bot] in #384
• Bump sentence-transformers from 5.1.1 to 5.1.2 by @dependabot[bot] in #385
• Bump fastmcp from 2.12.5 to 2.13.0.1 by @dependabot[bot] in #386
• Fix: default header mixed with session data. by @jealous in #398
• Fix ProfileMemory: improved prompt engineering and database schema fixes by @gauri-nagavkar-memverge in #374
• minor: set default empty to OPENAI_API_KEY by @jgong in #409
• Bump openai from 2.6.0 to 2.6.1 by @dependabot[bot] in #389
• Creating new Install Guide Section and adding AWS Bedrock and Ollama Support by @SarahScargall in #424

New Contributors

• @scottly06 made their first contribution in #341
• @tomw-mv made their first contribution in #231

Full Changelog: v0.1.7...v0.1.8

What's Changed

• add logging set to LOG_LEVEL by @ChristianKniep in #248
• Enable remote postgres and neo4j by @zzaym in #261
• Bump openai from 2.2.0 to 2.3.0 by @dependabot[bot] in #265
• Bump fastapi from 0.118.0 to 0.119.0 by @dependabot[bot] in #266
• Bump ruff from 0.13.2 to 0.14.0 by @dependabot[bot] in #264
• Bump pydantic from 2.11.10 to 2.12.0 by @dependabot[bot] in #263
• Bump neo4j from 5.28.2 to 6.0.2 by @dependabot[bot] in #222
• Use Issue Types rather than Labels by @sscargal in #281
• Support session data in the header. by @jealous in #245
• Unify the profile memory configuration by @malatewang in #258
• Revert "Create a new issue for helloworld.py example in Docker" by @sscargal in #292
• Revert HereDoc Syntax in Dockerfile by @sscargal in #280
• Have profile memory only use dependency injection by @o-love in #279
• Model config to use local model by @ChristianKniep in #284
• Fix LongTermMemory session deletion by @edwinyyyu in #273
• Docs: Edits to secure cfg.yml nomenclature and new unified cfg.yml file structure by @SarahScargall in #301
• Changing from G4 to GTM Container ID by @SarahScargall in #312
• Convert profile prompt to dataclass by @o-love in #300
• Add unit tests for profile memory by @o-love in #182
• Docs: Remove prompts from quickstart Guide, API path work by @SarahScargall in #313
• docs: improve episodic memory endpoint documentation by @dxu104 in #319
• memmachine-compose.sh fixes by @svetly-todorov in #306
• Docs: Fixing Instruction Set within QuickStart Guide by @SarahScargall in #314
• GTM Fix for Mintlify by @SarahScargall in #325
• memmachine-compose.sh: drop set -x by @svetly-todorov in #324
• Docs: Changing Capitalization from tagID to tagId by @SarahScargall in #327
• Docs: Docker-compose script --help update by @SarahScargall in #328
• Feature/add mcp stdio mode by @dxu104 in #323
• Add Amazon Bedrock embedder and reranker model support by @edwinyyyu in #146
• Add Amazon Bedrock language model support by @edwinyyyu in #271
• Update documentation_issue.yml by @sscargal in #337
• Bump fastmcp from 2.12.4 to 2.12.5 by @dependabot[bot] in #346
• Bump boto3 from 1.40.40 to 1.40.55 by @dependabot[bot] in #345
• Bump sqlalchemy from 2.0.43 to 2.0.44 by @dependabot[bot] in #344
• Bump uvicorn from 0.37.0 to 0.38.0 by @dependabot[bot] in #343
• Bump openai from 2.3.0 to 2.6.0 by @dependabot[bot] in #342
• [CI] Support Integration tests starting with ProfileMemory by @o-love in #318
• Fix error message "group_id or user_id cannot be empty" by @xiongzubiao in #333
• Remove envvar POSTGRES_PASS by @xiongzubiao in #332
• Validate OpenAI-compatible embedder dimensions by @edwinyyyu in #326
• Fix 'Terminated' errors of memmachine-compose.sh by @jgong in #356
• [CI] Add python 3.14 testing by @o-love in #353
• Docs: Update to configuration.mdx due to unification of config files, Amazon Bedrock Additions by @SarahScargall in #358
• Add inline document for REST API. by @jealous in #329
• Resolve Link issues and .cursor/rules.md file problem by @SarahScargall in #359

New Contributors

• @zzaym made their first contribution in #261
• @jealous made their first contribution in #245
• @dxu104 made their first contribution in #319
• @xiongzubiao made their first contribution in #333
• @jgong made their first contribution in #356

Full Changelog: v0.1.6...v0.1.7