12 projects
ragleap-vectorstores
Pluggable vector backends beyond ragleap-rag core's 6 (PgVector, FAISS, Pinecone, Weaviate, Qdrant, Milvus) - additional VectorBackend implementations as optional extras, no vendor lock-in, no bloated core dependency footprint.
ragleap-tools
Standalone, dependency-light tool implementations (calculator, file ops, datetime, unit conversion, JSON/CSV, text utilities) exposing OpenAI/Gemini-style function-calling schemas. No tool-calling execution loop - wire these into your own LLM tool-calling code, or into ragleap-agents when that ships. BYOK, no vendor lock-in.
ragleap-observability
Prometheus/Grafana/Loki/AlertManager observability Helm chart for RagLeap Core; the changelog states exactly what each release verified
ragleap-app-chart
Generic, reusable Helm chart for deploying arbitrary services to Kubernetes — not RagLeap-specific
ragleap-ops
Kubernetes deployment manifests for RagLeap Core, live-tested against a real cluster
ragleap-ansible
Ansible roles that prepare and harden a fresh Ubuntu server: user, sudo, updates, SSH and firewall
ragleap-terraform
Terraform module that creates a kind cluster and installs the RagLeap Helm charts, tested on a real cluster
ragleap-voice
Pre-alpha model manager for open, local speech-to-text and text-to-speech: licence-checked manifest, hash-verified downloads. No speech engines yet.
ragleap-agents
A small, provider-neutral agent loop over ragleap-tools Tool objects: a step cap, a taint rule that forces approval for outbound actions after untrusted content, pause/resume approval gates, and a per-run transcript. No database. Bring your own model as a plain callable, or use the included standard-library adapter for OpenAI-compatible chat APIs. BYOK, no vendor lock-in.
ragleap-integrations
MCP client that turns allowlisted tools on remote MCP servers into ragleap-tools Tool objects. Streamable HTTP only, standard library plus ragleap-tools, owner-configured allowlist, pinned and bounded network access. BYOK, no vendor lock-in.
ragleap-rag
Self-hosted RAG engine: hybrid dense+sparse retrieval, 28-format ingestion (docs, URLs, images, audio, video), 6 vector backends, 8 embedding providers, 12+ generation providers with automatic fallback, query rewriting (HyDE/contextual/multi-query), structured JSON output, real per-call cost tracking, and standalone retrieval via retrieve(). Composable with the optional ragleap-graph package for knowledge-graph-augmented retrieval. BYOK, no vendor lock-in, no hardcoded model defaults.
ragleap-graph
Knowledge-graph-augmented retrieval for RAG systems: Neo4j-backed entity extraction (regex or LLM-based), co-occurrence graphs, entity deduplication, and GraphRetriever for hybrid vector+graph retrieval via the optional ragleap-rag integration. Framework-agnostic, optional multi-tenant namespacing, no hardcoded models or vocabulary.
