knowledge-systems

Production-ready RAG framework for Python — multi-tenant chatbots with streaming, tool calling, agent mode (LangGraph), vector search (FAISS), and persistent MongoDB memory. Built on LangChain.

Lightweight, thread-safe memory kernel for autonomous agents. Strengthens knowledge through reinforcement, prunes with gentle decay, and exports transparent Markdown — pure SQLite. Foundation for persistent, self-improving reasoning.

Constitutional polis of AoA: ecosystem identity, layer map, federation rules, and program-level direction.

Building local and hybrid AI systems, agent workflows, and long-horizon knowledge architectures through AoA and Tree of Sophia.

A Streamlit platform that models academic research fields as market-like systems, with academic indices, AI interpretation, counterfactual analysis, and epistemic audits. 学术股市：用市场模型观察学术场的跨学科分析平台。

Evidence-backed engineering for knowledge-system harnesses: score, understand, improve, and verify.

Multi-agent autonomous research framework for AI-driven knowledge gathering, summarization, and fact verification workflows.

1956 Commons — an experimental AI-first collaborative environment governed by a constitutional framework and recorded through a repository ledger.

Production-style AI knowledge platform with FastAPI and Postgres/pgvector, enabling document ingestion, semantic retrieval, and grounded RAG APIs for real-world applications.

A modular governed framework for AI-assisted projects knowledge systems and data-governance research

Nepal-focused AI document understanding and grounded knowledge platform using Open Knowledge Format (OKF v0.2), offline embeddings, and Devanagari OCR.

Open research and coordination primitives for CDLI knowledge systems

Archived. Claude Code skills and playbooks published by Bamboo DCM, an independent structurer and distributor of corporate and structured credit in Brazil.

Public developer layer for CERVEL: experimental knowledge contracts, Python SDK, local validation, and a bounded localhost sandbox.

Bits Hierárquicos — um envelope estrutural que representa um ativo heterogêneo, navega por partes sem carregar tudo, e delega cada região ao especialista. Estudo medido (9 ângulos) + protótipo bhmem (memória de agente).