[Feat]: Implementation of Base Event-Driven Workflow

[smokeyScraper]

🚀 Feature Request

Implement the foundational event-driven system and workflow architecture for the Devr.AI bots.

📌 Description

We need to establish the core event-driven architecture that will serve as the backbone for all platform integrations. This system should allow for modular testing of individual bot components before integrating them into the unified workflow.

🎯 Targets

1. Event Bus System

• Define standardized event schema for cross-platform compatibility
  • Create base event class with common fields (platform, event_type, timestamp, actor)
  • Implement platform-specific event classes for GitHub, Discord, Slack, Discourse
• Create event producers for each platform
• Implement event consumers with appropriate routing logic
• Implement event serialization/deserialization with proper error handling

2. Event Handler Implementation

• Can be encapsulated as of now with a general class and base functions, later on to be extended as per individual bot's workflow which can be done via inheritance in individual bots.
  • IssueHandler - For GitHub issue events
  • PRHandler - For GitHub PR events
  • MessageHandler - For Discord/Slack messages
  • OnboardingHandler - For new contributor detection
  • FAQHandler - For knowledge-based questions
• Create handler registry system for dynamic event routing
• Fallback/Retry Mechanism

3. Vector DB Interactions

• Create a VectorDBService class for Supabase vector operations
• Implement EmbeddingGenerator service
• Design caching mechanism for frequent queries

4. Analytics Data Collection Updates

• Create event listeners for analytics events
  • Can be extended as per dashboard
• Pipeline for aggregating and storing events/updates (Preferably will redirect to the dashboard config part)

5. Event Processing Pipeline

• Implement EventEnricher service to add context to events
• Create EventPrioritizer to determine processing order
• Implement EventRouter to direct events to appropriate handlers
• Design ProcessingPipeline class with middleware support

6. Configuration Management

• Create configuration system for managing bot settings
• Implement feature flags for gradual rollout
• Add logging and monitoring hooks throughout the system

Initial start can be highly encapsulated with some tests of individual components preferably of directing/workflow

[smokeyScraper]

@chandansgowda, I have started aligning the base components and workflow. Can you please assign this to me? I am preferring here to make base classes and functions that can be later on extended to individual bots as per their requirements.
Thanks a lot!

[Vivekgupta008]

Hey @kartikbhtt7, I would love to collaborate with you on this. Have you done similar work before?

[smokeyScraper]

Hey @kartikbhtt7, I would love to collaborate with you on this.

yupp sure, as discussed we can surely collab on this part.

Have you done similar work before?

Yes, I have pretty good experience in working with event driven systems, was recently working on Paperflux, kind of event-driven (moreover scheduler scheduled up everyday for tasks aligned), currently extending this to interacting agents.

[Vivekgupta008]

That's great

[Vivekgupta008]

@chandansgowda Can you please assign this issue to us.

[muntaxir4]

Hey, since I am working on the RAG, I am already following OOPs principles. Can I work on the Vecror DB Integrations?

[Vivekgupta008]

Hey @muntaxir4, I have been collaborating with @kartikbhtt7 for this. Almost done with this.

[muntaxir4]

Hey @muntaxir4, I have been collaborating with @kartikbhtt7 for this. Almost done with this.

The RAG requires storing of files and quering the Vector DB, so I was also working on this.

[smokeyScraper]

Hi @muntaxir4 , we have already raise a PR for Vector DB integration part and seems to work good as of now. Can you please detail a bit more on storing of files.

The way we currently are implementing this that,
it's obvious there'll be some limited contextual texts/commands/references for any event so most probably they can be directly configured by the admin using the dashboard itself. This kind-of provides the admin a better control space and easy to make changes.
Have implemented the DB service rn and as of now working on Embedding Service using one the MTEB top benchmarked model, will be done probably by EOD. I think this will be good enough for the workflow as we're also having a fallback mechanism that is needed to be explored being directly relied on the LLM.

rest lmk we can have a discussion in the discord call/chat/group itself with the mentor guiding too.
my discord -> smokey_d_scraper

[muntaxir4]

Can you check the issue #17. I am already done with the RAG support, (it stores, retrieves files Supabase pgvector). Of course RAG requires embeddings hence I have used sentence transformer. And completed them on March 14 7:51pm. you can checkout the branch: https://github.com/muntaxir4/Devr.AI/tree/feat/RAG

[muntaxir4]

I will share the workarounds in afternoon