github-repo-hunter
自动发现和集成GitHub仓库的代理,为BidDeed.AI(房屋拍卖)和Life OS(生产力/ADHD)寻找相关的开源项目。搜索与域关键词匹配的仓库,通过README/技术栈分析评估相关性,自动将它们添加为目标仓库的子模块或存档,并向Ariel发送可操作摘要。当用户请求“查找相关GitHub仓库”、探索为BidDeed.AI工作流程的新集成、发现Life OS的生产力工具或保持仓库与最新开源创新同步时使用。
name: github-repo-hunter description: Autonomous GitHub repository discovery and integration agent that hunts relevant open-source projects for BidDeed.AI (foreclosure auctions) and Life OS (productivity/ADHD). Searches GitHub for repos matching domain keywords, evaluates relevance via README/tech stack analysis, auto-adds as submodules or archives to target repos, and alerts Ariel with actionable summaries. Use when (1) User requests "find relevant GitHub repos", (2) Exploring new integrations for BidDeed.AI workflows, (3) Discovering productivity tools for Life OS, (4) Keeping repositories updated with latest open-source innovations.
GitHub Repository Hunter
Autonomous agent for discovering, evaluating, and integrating relevant GitHub repositories into BidDeed.AI and Life OS ecosystems.
Mission
Hunt GitHub for open-source projects that could enhance:
- BidDeed.AI - Foreclosure auction intelligence (scrapers, ML models, document processing, workflow automation)
- Life OS - Productivity, ADHD management, swimming analytics, dual-timezone coordination
Core Workflow
1. Discovery Phase
Search GitHub API using domain-specific keywords.
Use scripts/hunt_repos.py with keywords from references/keyword_library.md.
2. Evaluation Phase
Score each discovered repo (1-100) using scripts/evaluate_repo.py.
Quick Filters (Auto-reject):
- Last commit >1 year ago
- <10 stars AND <5 forks
- No README
- Archived repository
Scoring Rubric:
- Tech Stack Match (30 pts) - Python/Rust/JS, Supabase, LangGraph, GitHub Actions
- Domain Relevance (40 pts) - Foreclosure tools, ADHD/productivity, swimming analytics
- Code Quality (20 pts) - Tests, docs, active issues
- Community (10 pts) - Stars, forks, contributors
Thresholds:
- Score ≥70 → AUTO_ADD to integrations/
- Score 50-69 → ALERT_ARIEL (needs review)
- Score <50 → SKIP (log to rejected_repos.txt)
3. Integration Phase
Execute scripts/integrate_repo.py for AUTO_ADD repos.
Integration Methods: A. Git submodule in integrations/ folder B. Reference entry in docs/integrations.md
4. Alert Phase
Insert discovery notification to Supabase insights table.
Alert Format:
🔍 GitHub Repo Hunter - Found: {repo_name}
Score: {score}/100 (AUTO_ADD / REVIEW_NEEDED)
Repository: https://github.com/{user}/{repo}
Language: {languages}
Last Updated: {last_commit}
Stats: ⭐ {stars} | 🍴 {forks} | 👥 {contributors}
What It Does:
{readme_summary}
Potential Use Cases:
- BidDeed.AI: {use_case}
- Life OS: {use_case}
Tech Stack Match:
✅ {matched_tech}
❌ {missing_tech}
Recommendation: {action}
Next Steps: {specific_action}
Scripts
scripts/hunt_repos.py - Main hunter with GitHub API
scripts/evaluate_repo.py - Scoring algorithm
scripts/integrate_repo.py - Auto-add to repos
References
references/keyword_library.md - Search keywords for both domains
references/github_api.md - GitHub Search API docs
references/scoring_rubric.md - Detailed evaluation criteria
Critical Rules
- Never auto-add without evaluation
- Respect GitHub API rate limits (5000 req/hr with auth)
- Test before integrating
- Document every integration in docs/integrations.md
- Alert on threshold (50-69 MUST notify Ariel)
Target Repositories
Primary:
breverdbidder/biddeed-conversational-aibreverdbidder/life-os
Secondary:
breverdbidder/brevard-bidder-scraper(archive-only)
Decision Tree
Discovered Repo
|
├─ Score ≥70? → Auto-add + Alert (FYI)
├─ Score 50-69? → Supabase alert + Review needed
└─ Score <50? → Skip (log rejection)