Community Sync - 2025

[xintongsong]

This page is outdated. Please find information about the latest community syncs at: #419

---

There's a community sync for discussing developing progress, technical designs and community issues about Flink Agents.

Everyone is welcome to join.

• Schedule: 3:30pm Wed UTC
• Google Calendar
  • The event might re-scheduled or skipped sometimes. It's strongly recommended to subscribe to the google calendar for update notifications.
• Google Meet

The agenda of the next sync and minutes of previous syncs will be posted below. Please feel free to propose new topics for the agenda.

---

2025-12-31

• Canceled

2025-12-24

• Canceled

2025-12-17

• Flink Agents 0.2
  • [Feature] Introduce Comprehensive Memory Management. #348 @wenjin272
  • Java API
    • [Feature][integration][Java] Java Implementation of Vector Store with Elasticsearch Support #321 @twosom
    • [Feature] [api][Java] Add built-in mcp support for Java #334 @xli1996 @yanand0909
  • [Feature] Support Cross-Language Resource #352 @GreatEugenius
    • Supports for the first Resource (ChatModel) is completed. Can use more help.
  • Support multiple Flink versions - Design WIP @GreatEugenius
  • Action Execution and Consistency - Design WIP, stale @Sxnan
• AI-Ops Demo @Sxnan

2025-12-10

• Flink Agents 0.2
  • [Feature] Introduce Comprehensive Memory Management. #348 @wenjin272
  • Java API
    • [Feature] [integration] [java] Integrate built-in Anthropic chat model #326 @xli1996
    • [Feature][integration][Java] Java Implementation of Vector Store with Elasticsearch Support #321 @twosom
    • [Feature] [api][Java] Add built-in mcp support for Java #334 @xli1996 @yanand0909
      • Require Java 17
      • CI: Add pipeline for Java 17, skip the tests for Java 11
      • Investigate multi Java version supports
  • [Feature] Support Cross-Language Resource #352 @GreatEugenius
    • Once the first Resource (ChatModel) is supported, we can use more help on other Resources.
  • Action Execution and Consistency - Design WIP, stale @Sxnan
  • Support multiple Flink versions - Investigating @GreatEugenius
• Flink Agents 0.1.1
  • Released

2025-12-3

• Flink Agents 0.2
  • Action Execution and Consistency - Design WIP @Sxnan
  • [Feature] Introduce Comprehensive Memory Management. #348 @wenjin272
  • Java API
    • [Feature] [integration] [java] Integrate built-in openai chat model #319 @xli1996
    • [Feature][integration][Java] Java Implementation of Vector Store with Elasticsearch Support #321 @twosom
    • [Feature] [integration] [java] Integrate built-in Anthropic chat model #326 @xli1996
    • [Feature] [api][Java] Add built-in mcp support for Java #334 @xli1996 @yanand0909
  • [Feature] Support Cross-Language Resource #352 @GreatEugenius
  • Need helps on
    • Subtasks of [Feature] Support Cross-Language Resource #352
    • Support multiple Flink versions
• Flink Agents 0.1.1
  • All bugfix completed, ready for the RC vote

2025-11-26

• Label-Based Version Control Strategy #302
  • All previous issues and PRs are labeled
• Flink Agents 0.2
  • Action Execution and Consistency - Design WIP @Sxnan
  • Comprehensive Memory Management for Flink-Agents #339 @wenjin272
  • Java API
    • [Feature] [integration] [java] Integrate built-in openai chat model #319 @xli1996
    • [Feature][integration][Java] Java Implementation of Vector Store with Elasticsearch Support #321 @twosom
    • [Feature] [integration] [java] Integrate built-in Anthropic chat model #326 @xli1996
    • [Feature] [api][Java] Add built-in mcp support for Java #334 @xli1996 @yanand0909
  • Resource Cross-Language Support for Flink Agents #338 @GreatEugenius
• Flink Agents 0.1.1
  • Issues
  • Try to fix the bugs in this week, and release in next week @wenjin272

2025-11-19

• Canceled

2025-11-12

• Flink Agents 0.2
  • Planning Flink Agents 0.2 #282
  • Progress
    • Comprehensive Memory Management: design wip
    • Cross-Language supports: design wip
    • Support fine-grained durable execution: design wip
    • Several bug fixes
      • [Bug] ReActAgent.stopAction() does not properly handle LLM output schema issue #274
      • [Bug] Python code under e2e_test cannot import the mcp module #276
      • [Bug] The test case fails to run on Windows #293
      • [Bug] PythonActionTask failed to serialize for Flink state #297
• Label-Based Version Control Strategy #302
  • Start enforcing it if not objection

2025-11-5

• Flink Agents 0.2
  • Planning Flink Agents 0.2 #282
  • Progress
    • Comprehensive Memory Management: design wip
    • Cross-Language supports: design wip
    • [Feature] Support ollama4j embedding models in java #304
    • [Feature] Azure LLM Integration for the Agent #286
  • Division of labor
    • We'll leave more time for people to go through the feature plan and to have offline discussions within their companies
• Flink Agents 0.1.1 (bugfix version)
• Label-Based Version Control Strategy #302
  • We'll leave it open for comments for a about 1 week, and if there's no objection we'll start to enforce it
• Feedback from Current

2025-10-29

• Canceled

2025-10-22

• Planning Flink Agents 0.2 #282
• Preparing Flink Agents 0.1.1 (bugfix release)
• Version control with GH Issues and Discussions
  • We are working on a solution based on the Pulsar label strategy. Once completed, we'll post in the community for discussion.
• Shall we move the community sync to a Europe-friendly slot?
  • Beijing - 11:30pm
  • Berlin - 5:30pm
  • California - 8:30am
  • We'll post the proposed slot in the Slack dev channel for more feedbacks

2025-10-15

• Canceled

2025-10-08

• Developing progress updates
  • Adding examples: completed
  • Bugfix and minor improvements: no known blocker issues
  • Documentation
    • All docs for Python is completed
    • Need to add instructions for Java
• Next: proceed with the release progress
  • Cut branch and create RC1 - today
  • Start voting - Oct 09~13
  • Polish docs and release announcement - During the vote
  • Finalize the release - Oct 14

2025-10-01

• Canceled

2025-09-24

• Developing progress updates
  • Phase 2: completed
  • Phase 3: completed
  • Adding example agents
    • python workflow example - completed
    • python react example - in progress @wenjin272
    • java workflow example - not started
    • java react example - not started
  • Packaging - completed
  • Documentation
    • Website setup: completed - https://nightlies.apache.org/flink/flink-agents-docs-main/
    • Overall structure: completed
    • Progress (0/13) - [Feature][docs] Flink-Agents MVP version project docs umbrella #196
  • Bugfix and minor improvements
    • Progress (5/7) - [Improvement] MVP user experience check #131

2025-09-17

• Developing progress updates
  • Phase 2
    • 1 remaining issue, @wenjin272 is working on it
  • Phase 3
    • MCP integration: PR will be merged today @yanand0909 @wenjin272
    • ChatModel integration: 4 providers on python side and 1 on java side are supported
    • Embedding Model: 2 providers on python side are supported
    • Vector Store: PR almost ready, blocked by the MCP integration PR @alnzng
    • Per-action state consistency: 2 PRs in review @letaoj
  • Phase 4
    • Website and documentation framework is ready, waiting for ASF to put it online
    • Documentation
      • @Sxnan is preparing the quick-start example agents and overall architecture of the user documentation
      • Flink Agent Project Docs #180
    • Bugs and issues found during preparing the example and documentation will be collected in the umbrella issue. If anyone has the capacity, please pick up and fix.
      • [Improvement] MVP user experience check #131

2025-09-10

• Developing progress updates
  • Phase 2
    • Python side all completed
    • Java side, 2 remaining issues, @wenjin272 will work on them
  • Phase 3
    • MCP integration: PR in review @yanand0909
    • EmbedModel & VectorStore: implementing @alnzng
    • ChatModel integration:
      • 4 providers / protocols are supported on python side: ollama, openai, anthropic, tongyi
      • We still need at least 1 supported on the java side, @letaoj will look into it after the per-action state consistency efforts
    • Per-action state consistency: 2 PRs in total, 1 is in review and 1 is being implemented @letaoj
  • Phase 4 (combined from previous phase 4 & 5)
    • @Sxnan is preparing the quick-start example agents and overall architecture of the user documentation
    • @GreatEugenius is preparing the website and documentation framework
• Time plan for the release
  • 9.22 - Major features completed, start testing, bug fixing and experience improving
  • 9.30 - User documentation completed, continue with bug fixing and experience improving
  • 10.9-14 - Vote and release

2025-09-03

• Canceled

2025-08-27

• Developing progress updates
  • Phase 2:
    • Java API for Prompt, Tool and ChatModel: PR in review @AmenAdem
    • ReAct Agent: PR in review @wenjin272
  • Phase 3:
    • MCP Integration: implementing @yanand0909
    • Built-In Chat Model
      • ollama: finished
      • openai: in-review @wenjin272
      • tongyi: in-review @KeGu-069
    • API & Abstraction for Embed Model & Vector Store: @alnzng will look into ti
    • Per-action state consistency: implementing @letaoj

2025-08-20

• Developing progress updates
  • Phase 2:
    • Java API for Prompt, Tool and ChatModel: PR in review @AmenAdem
    • ReAct Agent: In Progress @wenjin272
  • Phase 3:
    • MCP Integration: reach consensus on API, proceed with implementation @yanand0909
    • Built-In Chat Model
      • ollama: finished
      • openai: in-progress @wenjin272
      • tongyi: in-review @KeGu-069
    • Per-action state consistency: design in discussion @letaoj
    • Event log and listener: completed

2025-08-13

• Developing progress updates
  • Phase 1: Completed
  • Phase 2: 2 Remaining issues, both being worked on
  • Phase 3:
    • Design / codes being discussed
      • MCP Server (Tools, Prompts) @lihaosky
      • Per-action state consistency @letaoj
      • Event-log & event-listener - @alnzng
    • Completed:
      • Metrics mechanism
      • Async execution in python
      • Ollama chat model integration
• We went through the Per-Action Consistency Design from @letaoj

2025-08-06

• Developing progress updates
  • Phase 1: 1 remaining issue, PR in review.
  • Phase 2: Mostly done on the python side, only 1 remaining issue. Java side is still being worked on.
  • Phase 3: Design of the following items are in progress
    • MCP Server (Tools, Prompts), EmbedModel, VectorStore @lihaosky
    • Per-action state consistency @letaoj
    • Event-log & event-listener - @alnzng
• We went through the Memory Design from @coderplay.

2025-07-30

• Developing progress updates
  • Phase 1: Almost done. Only 1 remaining issue, @xli1996 and @letaoj will look into it.
  • Phase 2: Still in progress. @wenjin272 is working on the Python side. @lihaosky will find people to work on the Java side.
• Time-plan review
  • We are 2 weeks behind the schedule
  • So we decide to downgrade the following items in Phase 3 as optional for the MVP version
    • Hierarchy Agent
    • Cross-language invocation
    • Local python runtime
  • We also assigned some important items in Phase 3 as follows. The assignee would create an initial design doc for further discussions.
    • MCP Server (Tools, Prompts), EmbedModel, VectorStore - @lihaosky
    • Long-term memory - @coderplay
      • @coderplay shared a wip design and will post in Github discussion once completed.
    • Per-action state consistency - @letaoj
    • Event-log & event-listener - @alnzng
• We aligned around the timeline targeting a MVP release by the end of September, which means most of the development should be completed in early September.
  • @thefalc will post the target use cases of the MVP release in Github discussion

2025-07-24

• Canceled