Deodat-Lawson/LaunchStack

Implements a three-layer modular architecture (Services, Tools, Physical) with domain-partitioned data isolation enforced via Clerk RBAC, enabling vertical business modules (Marketing, Legal, Onboarding) to share reusable AI tools like RAG, web search, and entity extraction. Supports importing knowledge from Notion, Google Workspace, Slack, and GitHub via standardized adapters, with domain-specific document analysis that auto-detects missing artifacts (contracts, financials, compliance docs) and extracts actionable insights using chain-of-verification and supervisor agent validation. Built on LangChain agent guardrails (PII filtering, grounding checks, confidence gating) with optional observability via Inngest and LangSmith, plus social media content generation pipelines targeting Reddit, X, LinkedIn, and Bluesky.

788 stars. Actively maintained with 91 commits in the last 30 days.