Turn Any Codebase Into
Pre-Indexed Knowledge for AI Agents.
LangGraph multi-agent pipeline extracts deep database schemas, API contracts, and 6-file technology modules. Automatically delivers repowiki/, CLAUDE.md, and AGENTS.md straight to your GitHub PR.
$ saaya generate --repo=https://github.com/owner/repository
➜ [1/5] Universal Smart Code Scanner: Discovered 240 files & sampled 18 schemas/controllers
➜ [2/5] Tech-Stack Extractor: Identified 18 topics (PostgreSQL, Redis, Better Auth, BullMQ, Pusher)
➜ [3/5] Multi-Pass Domain Planner: Planned 72 nested sub-articles across 8 functional domains
➜ [4/5] Deep Article Writer: Generated Mermaid sequence diagrams & code contract citations
➜ [5/5] Knowledge Graph Linker: Connected 82 relations in repowiki-metadata.json
✓ SUCCESS: Pull Request created on owner/repository!
PR Link: https://github.com/owner/repository/pull/42
Pre-Indexed Accuracy
Zero context hallucination for AI coding agents
Nested Docs per Scan
Hierarchical sub-articles & 6-file module cards
Token Window Savings
Instant context maps for Claude Code & Cursor
GitHub PR Delivery
Automated tree commits without manual writing
Engineered for Architectural Excellence
Every component in Saaya is designed to extract deep codebase understanding and feed it cleanly to human engineers and AI coding agents.
LangGraph Multi-Agent Pipeline
Stateful execution graph orchestrating specialized AI nodes: planTaxonomy $\rightarrow$ generateCatalogs $\rightarrow$ generateModuleCards $\rightarrow$ writeArticles.
PASS 1
Taxonomy
PASS 2
Catalogs
PASS 3
Tech Suites
PASS 4
Graph Link
CLAUDE.md Auto-Discovery
Includes root CLAUDE.md and AGENTS.md guidelines. Claude Code, Antigravity, and Cursor auto-read repowiki maps upon load.
Dual Knowledge Structure
Human-readable sub-documents with Mermaid sequence diagrams + machine-parseable 6-file tech cards (_module.yaml).
Tech-Stack Auto-Discovery (20+ Modules)
Scans code imports and dependencies to identify databases, ORMs, queues, auth, and AI models. Builds 6-file documentation suites for each.
From Codebase to Knowledge Base
A fully autonomous 4-stage pipeline that runs without manual intervention.
Universal Codebase Ingestion
Paste any public or private GitHub repository URL. Saaya automatically scans directory trees, ORM schemas, route handlers, async workers, and configurations across any language.
➜ saaya scan https://github.com/org/repo [Scanner] Discovered 240 files across 12 packages [Scanner] Sampled 18 core schemas & controllers
Insanely Detailed Generated Output
Inspect live examples of the articles, AI agent instructions, and knowledge graph JSON delivered to your repository.
# AI Agent Guidance & Architectural Map
This repository includes a pre-indexed knowledge base under repowiki/:
## Essential Context References for AI Agents (Claude Code, Antigravity, Cursor, Aider)
Whenever working on features, bug fixes, or refactoring, inspect these maps first:
- 🏗️ System Architecture: repowiki/en/content/Architecture Overview/System Architecture.md
- 🗄️ Database Schemas: repowiki/en/content/Database Schema/Database Schema.md
- 🔌 Backend API Services: repowiki/en/content/Backend API/Backend API.md
- ⚡ Background Workers: repowiki/en/content/Background Workers/Worker Architecture.md
## Coding Conventions & Execution Commands
Refer to 6-file module suites under repowiki/knowledge/en/:
1. coding_conventions.md — Code style, error handling, layering
2. unique_setup_and_commands.md — Build, run, and test execution commandsTransform Your Codebase in 60 Seconds.
Paste any public or private GitHub repository URL below to generate a pre-indexed Repowiki and AI agent context package.