Building RepoRank — The Open Source Impact & Funding Readiness Engine

Описание и применение

RepoRank объединяет метрики из GitHub, PyPI, npm, HackerNews и Open Collective с помощью Coral SQL, а затем использует LLM для генерации оценки влияния, описания и рекомендаций по грантам, ускоряя процесс подготовки заявок на финансирование.

Задача

Maintainers struggle to compile impact metrics from multiple APIs to craft grant proposals and sponsorship applications.

Результат

RepoRank reduces data aggregation from days to seconds, produces a 0‑100 impact score, a concise pitch, and targeted grant suggestions, streamlining funding workflows for open‑source maintainers.

Рабочий процесс

• Define custom YAML source specs for PyPI, npm, HackerNews, Open Collective
• Integrate sources into Coral SQL engine
• Write a single cross-source JOIN query to fetch metrics
• Execute query via Coral to obtain unified JSON
• Pass JSON to Hugging Face Qwen 2.5 (72B) for analysis
• Generate impact score, narrative, pitch, and grant recommendations
• Display results in a dark‑mode dashboard with radar charts and grant cards

Доказательства

Integrating 4+ external APIs usually takes days of reading docs and writing API wrapper code.

Источник

Coral changes this completely by letting you query APIs using standard SQL.

Источник

Using structured JSON schema outputs, Qwen acts as an expert open-source analyst, returning Impact Score, Narrative, Pitch, and Targeted Grants.

Источник