Core RAG pipeline tools - universal chunking, embedding, vector store interfaces
Defend an agent against hidden-content injection: strip payload-capable invisible Unicode and ANSI, splice out human-invisible HTML, and flag data-exfil URLs in untrusted text before any model sees it.
V3 PDF intelligence MCP server with smart Agent Document Twin extraction, focused evidence operations, OCR provenance, trust/accessibility reports, and benchmark-gated release proof.
Constella — run autonomous AI agent-companies locally: real claude/codex agents, Goals → Specs → Issues → Plans, local models, GitHub & Telegram, all from one cosmic control plane.
CLI for the Virage RAG pipeline — provides the virage command
Local transformers.js embedding provider for @vivantel/rag-core
LanceDB vector store for @vivantel/rag-core (embedded, file-based)
Full Court Defense CLI — security scanning for AI agents from your terminal
PostgreSQL + pgvector store for @vivantel/rag-core
MCP stdio server for Virage RAG — search and inspect your index from AI assistants
OpenAI-compatible embedding provider for @vivantel/rag-core
ChromaDB vector store for @vivantel/rag-core (local or hosted)
Qdrant vector store for @vivantel/rag-core (local and cloud)
Built-in chunking strategies for @vivantel/rag-core
Local cross-encoder re-ranker for Virage (ONNX, no API key required)
AST-aware code chunking strategy for @vivantel/virage-core, powered by code-chunk
Local-first memory MCP server for Claude Code and any MCP client: hybrid vector + knowledge-graph search, bi-temporal recall, Obsidian sync. Runs 100% on your machine, $0/token.
Mr.Sliy - 基于Tree-sitter与RAG的多语言代码优化智能体
Full AI context layer over MCP — tree-sitter code-map, document RAG (PDF/Office/HTML/email + OCR + reranker), shared agent memory, on-demand web crawl, git history + blame + per-symbol diff. 300+ languages, 10+ coding-agent harnesses, content-addressed Fj
AI coding assistant skill (Claude Code, Codex, Gemini CLI, Kimi Code, GitHub Copilot CLI, Aider, OpenCode, OpenClaw) - turn any folder of code, docs, papers, images, or audio/video transcripts into a queryable knowledge graph