kai2ser/pefa-rag
AI-powered semantic search across PEFA (Public Expenditure and Financial Accountability) national assessment reports. Next.js + Supabase pgvector + Anthropic Claude.
SUMMARY AI summary by gpt-5-mini
PEFA Reports AI is a Next.js application that provides AI-powered semantic search over PEFA (Public Expenditure and Financial Accountability) national assessment reports. It’s intended for analysts, practitioners, and teams who need to find and query content across a curated collection of PEFA reports. Key features: - Semantic search via OpenAI embeddings (text-embedding-3-large) stored in Supabase Postgres with pgvector. - Natural-language responses generated/streamed by Anthropic Claude Sonnet 4. - Shared documents catalogue (CSV → Supabase documents table) scoped to collection_id = "pefa". - Data pipeline: seed overview rows, download public PDFs, parse/chunk/embed, upsert vectors to pefa_chunks. - Deployment: Next.js 16 on Vercel; requires five env vars and a Supabase migration. Includes dev/build/lint scripts and an estimated embedding cost for full ingest.
AI Coding Tools Used
This repository contains config files from these AI coding tools, indicating AI-assisted development.
DETECTED Detected AI stack
AI-related keywords found in this repo's description, topics, or README summary — grouped by category. Each badge links to the corresponding ranking detail page.
Language breakdown (by bytes)
Owner
Senior Economist, but with an interest bringing together people, processes and technology for better public sector performance and development outcomes.
Dates
| Created on GitHub | 2026-05-16 |
| Last push | 2026-05-16 |
| First seen here | 2026-05-16 |
| Last fetched | 2026-05-16 18:11 |
Similar repos (same language)
Claude Code 桌宠伴侣 — 透明桌宠窗口实时显示 Claude Code 的工具调用、会话状态和完成提醒。
YieldDi/cc-devloopClaude Code Dev Loop
ArisRhiannon/vibecheckOffline, no-AI "safe to ship?" gate for vibe-coded apps — finds the security mistakes AI coding agents make (secrets, unprotected routes, eval-RCE, SQLi, CORS, JWT, leaked keys). CLI + MCP, agent-runnable.
neang-mengseang/LLM-RouterSelf-hosted OpenAI-compatible API that routes LLM requests across 14 providers with automatic failover.
civilization-os/CodeCartographerStatic code intelligence for generating architecture docs, call graphs, business flows, and structured repository maps with optional LLM-powered method understanding.
nithiin7/repo-chatA simple and powerful RAG chatbot that can index and intelligently answer questions about your codebase.
code-by-tiago/PlanifyPlataforma de Automação Educacional com IA Repositório oficial de uma aplicação web externa para o ecossistema Planify. O sistema utiliza modelos de linguagem (LLMs) para processar diretrizes pedagógicas, cronogramas estruturados de ensino automatizados e exportar materiais didáticos formatados (PDF/Docs) para educadores.
boggspa/TaskWraithLocal-first macOS workbench for running and reviewing AI coding agents — multi-provider, with multi-agent ensembles, diff review, and on-device history.