Deepfunding-Agent

Build AI Agent to better funding public goods, specifically for deepfunding.org

Usage

1. Install Poetry for Python dependency management.
2. Prepare .env

poetry install
cp .env.example .env

Huggingface part1 python notebook and partial result

How It Works

1. Metrics Collector

• Gathers repository metrics from OSO
• Fetch README content
• Reasoning and Plan what to search online for additional information
• Send all metrics to Analyzer Agents

2. Multi Analyzer Agents

• Involving three experts: Project Analyzer, Funding Strategist, and Community Advocate.
• Each agent reads the metrics and provides structured analysis including weights, reasoning, and confidence.
• Project Analyzer: Evaluates technical aspects and project fundamentals
• Funding Strategist: Focuses on funding history and resource allocation
• Community Advocate: Analyzes community engagement and ecosystem impact

3. Validator: Check each agent's analysis is comprehensive and well-justified. If not, send back to analyzer agent for revision.
4. Consensus: Combine results from all agents, calculate final weights

Example Output

Multi Analyzer Agents

Project Analyzer: {
  "weight_a": 0.4,
  "weight_b": 0.6,
  "confidence": 0.9,
  "reasoning": "Repo A (WalletConnect) has a high number of active developers (30) and significant recent activity with over 1954 commits in the last 6 months. However, it has a large number of open issues (250) and closed issues (212), indicating potential sustainability issues and a backlog of user concerns. Despite the promising metrics, the lack of funding in the last 6 months raises red flags about future support. Meanwhile, Repo B (Tokio) has a solid star count (86203) and a more extensive contributor base (107). Although its recent activity metrics are lower than Repo A, it has received funding recently and has a robust history and community backing, suggesting strong potential for future sustainability. The relative metrics indicate that while Repo A shows immediate activity, Repo B has stronger community and financial backing, which suggests long-term viability. Given these factors, I assign a weight of 0.4 to Repo A and 0.6 to Repo B.",
  "metrics_used": [
    "activeDeveloperCount6Months",
    "closedIssueCount6Months",
    "commitCount6Months",
    "contributorCount",
    "forkCount",
    "starCount",
    "total_funding_received_usd",
    "newContributorCount6Months",
    "openedIssueCount6Months"
  ]
}

Validator

{"validation": "The analyses provided by all three analyzers are comprehensive, well-justified, and actionable. Each analysis covers project health metrics (star count, fork count, dependents), funding strategy (total funding), and community impact (dependency rank). They all arrive at the conclusion that Repo A deserves a higher weight due to its superior metrics. The reasoning includes concrete metrics and logical deductions supporting their claims. However, the Community Advocate's analysis could benefit from a bit more detail on how Repo B's niche role might affect future contributions or funding. Therefore, the Community Advocate should revisit their analysis to enhance clarity in that aspect.", "revision_needed": "community_advocate"}

Consensus

{
  "weight_a": 0.44648544059324285,
  "weight_b": 0.5535145594067573,
  "confidence": 0.9,
  "agent_influence": {
    "project_analyzer": 0.33215368880442675,
    "funding_strategist": 0.33960699996742183,
    "community_advocate": 0.3282393112281511
  },
  "metric_importance": {
    "activeDeveloperCount6Months": 0.3213927660247748,
    "closedIssueCount6Months": 0.3213927660247748,
    "commitCount6Months": 0.3213927660247748,
    "contributorCount": 0.3213927660247748,
    "forkCount": 0.3213927660247748,
    "starCount": 0.3213927660247748,
    "total_funding_received_usd": 0.3213927660247748,
    "newContributorCount6Months": 0.22332407749776817,
    "openedIssueCount6Months": 0.3213927660247748,
    "mergedPullRequestCount6Months": 0.23202309793729778,
    "openedPullRequestCount6Months": 0.23202309793729778,
    "contributorCount6Months": 0.12692690822864902
  }
}

Development

Add/Change Agents

langchain tutorial

workflow.add_node in deepfunding.py

VsCode

Get the poetry environment path and paste it to your VsCode Python: Select Interpreter

poetry env info --path | pbcopy

Poetry

Add new dependencies

poetry add <package>