# Catalogue

> Public catalogue components.

JSON: https://leyline.ahensya.com/api/v1/components

- [LlamaIndex](https://leyline.ahensya.com/components/llamaindex.md): Data framework for retrieval and agents. Memory and index slices, not a whole 'RAG product' box.
- [PageIndex](https://leyline.ahensya.com/components/pageindex.md): PageIndex (MIT, Python) is VectifyAI’s vectorless, reasoning-based RAG engine. First-party pitch: build a hierarchical tree index per document, then let an LLM reason through that tree (human-like section navigation) instead of embedding chunks into a vector database.
- [Casdoor](https://leyline.ahensya.com/components/casdoor.md): An open-source, self-hosted identity and access management platform and authentication server with a web console, supporting SSO, multiple identity protocols, policy-based authorization, APIs, SDKs, and MCP.
- [cognee](https://leyline.ahensya.com/components/cognee.md): Open-source memory system for agents and teams. It connects company sources such as Slack, GitHub, and Linear so agents can recall shared context through one memory API, an SDK, or an MCP server, with an optional Cognee Cloud for managed scale.
- [fx](https://leyline.ahensya.com/components/fx.md): fx is a Vercel Labs coding-agent harness and CLI written in Zig. Official copy positions it as a tiny, open, native agent originally used internally, now open sourced for research and embedding in larger systems. The durable object is the product at fx.sh / vercel-labs/fx, not the tweet.
- [memini](https://leyline.ahensya.com/components/memini.md): Self-hosted persistent memory service for MCP-capable agents. It stores and recalls memories in a tiered store using hybrid vector and keyword retrieval, ships as a single Go binary with SQLite or Postgres, and exposes REST, MCP, and an embedded admin UI.
- [Self](https://leyline.ahensya.com/components/self.md): Self-Hosted AIEnterprise Search18 min readPublished May 22, 2026
- [turbopuffer](https://leyline.ahensya.com/components/turbopuffer.md): turbopuffer is a managed, commercial search engine from turbopuffer Inc. (Canadian, founded 2023). It is object-storage-first: Amazon S3 / Google Cloud Storage / Azure Blob is the source of truth; NVMe and memory are caches. It does approximate vector search (SPFresh), BM25 full-text, filters, and hybrid fusion.
- [Alibaba Cloud Model Studio](https://leyline.ahensya.com/components/alibaba-model-studio.md): Model APIs for text, vision, speech and embeddings, with compatible developer interfaces.
- [Local FAISS MCP Server](https://leyline.ahensya.com/components/local-faiss-mcp-server.md): Local FAISS vector store exposed as an MCP server for document ingestion, embeddings, persistent indexes, semantic search, optional re-ranking, and RAG-style retrieval with MCP prompts.
- [Octen](https://leyline.ahensya.com/components/octen.md): Search infrastructure for web, image and video retrieval, with related embedding and research capabilities.
- [pen.dev](https://leyline.ahensya.com/components/pen-dev.md): Pencil / pen.dev (site banner: “pencil.dev is now pen.dev”) is an AI-native infinite design canvas positioned as “Dream on canvas. Land in code.” Backed by a16z Speedrun (site badge). Integrates with Cursor, VS Code, Claude Code, OpenAI Codex; emphasizes bi-directional MCP vector canvas, open file format, Figma bring-over, brand kits, and design-as-code.
- [QuiverAI](https://leyline.ahensya.com/components/quiverai.md): QuiverAI ships foundational models for generating, editing, and animating editable vector/SVG assets (logos, illustrations, diagrams, line work; typography & animations marked coming soon on marketing site). Arrow 2 is the Sep 2026 model line called out as higher quality and faster, in App and API.
- [Needle 3](https://leyline.ahensya.com/components/needle-3.md): Needle 3 is Cactus Compute’s on-device “automation foundation model”: a laddered / sliceable tiny model (post: 8–29 MB CQ2-bit binary; 25–121M params; depths 2–20 layers from one weight set) built on their Simple Attention Networks, aimed at tool calls, structured extraction, and local text embeddings — not open-ended chat.