oss-veda

oss-veda

Wisdom for your next open source contribution.

A Claude Code plugin that scouts trending AI/ML repositories across 5 data sources in parallel, then ranks contribution opportunities by career impact -- combining repo momentum, issue quality, and your personal skill fit.

---

Table of Contents

• Install
• Usage
• How It Works
• Architecture
• Scoring Algorithm
• Configuration
• Data Sources
• Skills & Commands
• Requirements
• Troubleshooting
• License

---

Install

# Add the marketplace
/plugin marketplace add anujdevsingh/anuj-ai-tools

# Install oss-veda
/plugin install oss-veda@anuj-ai-tools

Usage

Quick scan (7-day lookback)

/veda

Deep scan (14-day lookback, more repos)

/veda-deep

Natural language

Just ask Claude any of these:

• "What should I contribute to this week?"
• "Find me something to contribute to"
• "What's hot in AI right now?"
• "Show me good first issues in LLM repos"

Analyze a specific issue

Paste a GitHub issue URL and ask: "Should I work on this?" This triggers the veda-rank skill for deep-dive analysis.

Draft a contribution

After finding an issue, say: "Help me contribute to this" This triggers the veda-write skill to draft the PR.

---

How It Works

                         /veda
                           |
                    run_scouts.py
                    (orchestrator)
                           |
            +--------------+--------------+
            |       |       |      |      |
         GitHub    HN    Reddit   HF    PwC
         scout   scout   scout  scout  scout
            |       |       |      |      |
            +--------------+--------------+
                           |
                  oss-veda-raw.json
                           |
                 rank_opportunities.py
                   (scoring engine)
                           |
                 oss-veda-ranked.json
                           |
                  generate_report.py
                           |
                 Markdown report + strategic pick

Pipeline steps

1. Scout (parallel) -- 5 async scouts fetch trending repos, posts, models, and papers from GitHub, Hacker News, Reddit, HuggingFace, and Papers with Code simultaneously using asyncio.gather().

2. Rank -- Each repo+issue pair is scored on 3 axes (repo momentum, issue quality, skill fit) producing a composite career impact score out of 100.

3. Report -- Top 20 opportunities are formatted into a markdown report with executive summary, ranked table, and long-shot picks.

4. Strategy -- Claude adds a personalized "This Week's Strategic Pick" based on your skill profile and current AI hiring trends.

---

Architecture

oss-veda/
|-- .claude-plugin/
|   |-- plugin.json                    # Plugin manifest
|-- skills/
|   |-- oss-veda/
|   |   |-- SKILL.md                   # Main skill: find opportunities
|   |   |-- references/
|   |       |-- api-endpoints.md       # API endpoint reference
|   |       |-- scoring.md             # Scoring algorithm docs
|   |       |-- skill-profile.md       # Default user profile
|   |-- veda-rank/
|   |   |-- SKILL.md                   # Deep-dive issue analyzer
|   |-- veda-write/
|       |-- SKILL.md                   # Contribution drafter
|-- agents/
|   |-- github-scout.md               # GitHub trending repos + issues
|   |-- hn-scout.md                    # Hacker News Show HN posts
|   |-- reddit-scout.md               # Reddit AI subreddits
|   |-- hf-scout.md                   # HuggingFace trending models/spaces
|   |-- pwc-scout.md                  # Papers with Code linked repos
|-- commands/
|   |-- veda.md                        # /veda slash command
|   |-- veda-deep.md                   # /veda-deep slash command
|-- scripts/
|   |-- run_scouts.py                  # Orchestrator (parallel execution)
|   |-- github_scout.py               # GitHub Search API client
|   |-- hn_scout.py                    # HN Algolia API client
|   |-- reddit_scout.py               # Reddit public JSON client
|   |-- hf_scout.py                    # HuggingFace API client
|   |-- pwc_scout.py                   # Papers with Code API client
|   |-- rank_opportunities.py          # Career impact scoring engine
|   |-- generate_report.py             # Markdown report generator
|-- config.json                        # Centralized configuration
|-- README.md

Key design decisions

• uv single-file scripts -- Each Python script uses PEP 723 inline metadata (# /// script). Dependencies are declared per-file and auto-installed by uv on first run. No global requirements.txt, no shared virtual environment.