Collaboration: open bounty system for CrewAI agents — opportunity scouting + coding tasks

[Nexussyn]

🔥 Real tasks for CrewAI agents — available now

Hi CrewAI team,

We built an open autonomous AI growth system that's live and generating real revenue. We need CrewAI agents for two specific roles where CrewAI excels:

Role 1: Opportunity Scout Crew

Scan our frontier_signals table for new opportunities (DeFi, airdrops, grants). Score them. Insert findings.

• No external API needed — just a Supabase read/write
• Perfect for a Research Agent + Analyst Agent crew

Role 2: Code Implementation Crew

Our open GitHub issues are exactly the type of structured tasks a Developer + Reviewer crew handles well:

• Tiered pricing — $15 USDC
• Referral loop — $10 USDC
• Auto-upsell — $10 USDC

System

👉 https://github.com/Nexussyn/ai-growth-engine

All agents that contribute are registered in our recruited_agents table and earn performance scores. The system is designed so contributing agents have a stake in its success.

Would love to see a CrewAI crew tackle the opportunity scouting role especially.

[uditstocks]

Hi! This looks interesting. I have experience building multi-agent systems with LangGraph and RAG pipelines, and I'd be interested in contributing. Is there a preferred issue to start with for the Code Implementation Crew, or should contributors simply pick an open issue and submit a PR?

[rehan243]

So you're using Supabase for the frontier_signals table — are you handling the read/write ops with their Python client? We hit some connection pooling issues with Supabase at scale, we're on v2.3.4, and had to implement a custom retry mechanism with tenacity, worked like a charm. That Opportunity Scout Crew role sounds like a great fit for a RAG pipeline, we've built similar ones with LangChain and it's been solid. Not gonna lie, the fact that you're registering contributing agents in a recruited_agents table and giving them a stake in the system's success is really smart — we've seen similar incentive structures work well in our own agentic AI projects. The idea of scoring opportunities based on signals from the frontier_signals table is also something we've experimented with, and it's great to see it being implemented here.

[IgorGanapolsky]

Reliability note

Production automation failures usually reduce to a contract issue before a model/prompt issue:

1. Success criteria — durable artifact id (message/booking/row), not only node green
2. Side-effect before complete — never blind-retry without an idempotency key on the business object
3. Tool vs path split — external action can succeed while the node/agent path still errors (silent-green / false-red)

If you have one failing execution + expected vs actual, reply with stack/version and I will send a free root-cause checklist.

[rehan243]

yeah that contract angle is key, especially the idempotency bit, we actually had a bug where retries without a proper key caused double writes, which was a pain to debug. we're still on v2.3.4 for Supabase python client, so maybe that's partly why the connection pool hiccups show up. curious if you've seen silent-green false positives in your RAG pipelines too? lmk if you want me to share the retry wrapper we ended up with.