-
Notifications
You must be signed in to change notification settings - Fork 3
Home
Welcome to the ragleap-core monorepo wiki. This repo hosts the open,
pip-installable layer of the RagLeap ecosystem: focused, independently
versioned Python libraries published to PyPI, plus the standalone
self-hostable app they're built alongside.
If you're looking for the commercial multi-tenant SaaS platform
(ragleap.com), that's a separate, private repo — this wiki and
packages.ragleap.com cover the open-source packages only.
| Package | Purpose | PyPI version | Tests | Docs |
|---|---|---|---|---|
ragleap-rag |
Core retrieval — 6 vector backends, guardrails, query rewriting, structured output, cost tracking | v0.12.3 | 245 passing | packages.ragleap.com |
ragleap-graph |
Neo4j knowledge-graph retrieval — entity/relation extraction, lineage tracing, RBAC, audit logging, concurrent-write safety | v0.6.7 | 89 (88 passed, 1 skipped w/o GEMINI_API_KEY) |
packages.ragleap.com |
Status definitions: Live = published to PyPI today · Next = actively planned, not yet started · After = planned, depends on the "Next" tier landing first · Ongoing = grows continuously alongside every phase · Decision needed / Later = intentionally not yet scoped · Vision = early discussion only, explicitly not a committed roadmap item.
| # | Package | Purpose | Status |
|---|---|---|---|
| 1 | ragleap-rag |
Core retrieval | Live — v0.12.3 |
| 2 | ragleap-graph |
Neo4j knowledge-graph retrieval | Live — v0.6.7 |
| 3 | ragleap-vectorstores |
Pluggable vector backends beyond ragleap-rag core's 6 |
Next |
| 4 | ragleap-tools |
Built-in tools: search, code exec, calculators, file ops | Next |
| 5 | ragleap-integrations |
MCP-native connectors + curated native integrations, building on existing channel code | Next |
| 6 | ragleap-agents |
Role-based crews, tool-calling, HITL approval gates | After |
| 7 | ragleap-flows |
Low-code orchestration + HITL, bootstrapped from existing n8n integration | After |
| 8 | ragleap-observability |
Tracing, hallucination detection, LLM-as-judge eval framework | Decision needed — no internal precedent yet |
| 9 | ragleap-ops |
Docker/K8s templates, Helm charts, CI/CD, IaC | Ongoing |
| 10 | ragleap-studio |
Visual low-code builder UI | Later |
| 11 | ragleap-memory |
Shared long-term memory layer across agents/employees | Later — an untracked scaffold directory exists from concurrent automation, not yet a scoped package |
| 12 | ragleap-intelligence |
"AI Employee Twin" vision — persistent per-employee AI assistant | Vision only — not scoped, not committed |
For the full narrative behind each phase, see Roadmap.
| Layer | What | Where |
|---|---|---|
| 1. Commercial platform | Multi-tenant SaaS, AI Employees, billing, managed hosting |
ragleap.com, private ragleap-backend repo |
| 2. Standalone open-source app | Full "RagLeap Core" app — FastAPI web layer, WhatsApp/Telegram/Discord/Voice channels, Neo4j knowledge graph, docker compose up deployable |
This repo — core/, channels/, db/
|
| 3. Focused pip-installable libraries |
ragleap-rag, ragleap-graph (live today), 10 more planned |
This repo — packages/
|
docs.ragleap.com covers Layer 1 only. If you're here for the pip
packages, you're in the right place — see packages.ragleap.com for API
reference and quickstarts.
- Every claim in code, docs, and marketing is either live-verified against real infrastructure or explicitly labeled unverified. No projected or estimated numbers presented as measured.
- PRs go through the standard 14-step release cycle for anything touching a published package (see Roadmap for the full checklist).
- Open issues are the source of truth for what's actually planned next — check the issue tracker before proposing new scope, especially around items marked "decision needed" above.
- This monorepo has concurrent automated sessions working on it alongside human contributors. If you see unfamiliar branches or stashes, they're very likely legitimate in-progress work — please don't discard them.
Last synced with Roadmap.md and PyPI as of ragleap-graph v0.6.7 /
ragleap-rag v0.12.3. If anything here looks stale, the PyPI version
pages and packages.ragleap.com are the live source of truth.