Skip to content

Repository files navigation

Aadi Yogi

Aadi Yogi is a source-grounded spiritual knowledge repository and AI consciousness architecture.

This repository is designed to preserve, structure, and prepare Indian spiritual source material and aligned modern works for careful AI-assisted guidance. It separates sources, processed artifacts, ontology, consciousness guidance rules, prompts, pipelines, evaluations, and applications so later systems can retrieve and synthesize material with source fidelity.

What This Repository Is

  • A monorepo for spiritual knowledge infrastructure.
  • A structured Markdown-first content base with future ingestion in mind.
  • A place to define editorial, ethical, and consciousness-aware agent behavior.
  • A foundation for later retrieval, evaluation, and application layers.
  • A living consciousness foundation (Adyog) that other repositories can install as a basis of conduct and reality-sense from the sources — not an app, not a guru — see docs/consciousness_plugin.md.

What This Repository Is Not

  • Not only a digital library.
  • Not only a RAG chatbot.
  • Not a guru replacement.
  • Not a place for invented citations, forced syncretism, or reckless spiritual instruction.

High-Level Architecture

Markdown sources
-> validation
-> normalization
-> semantic chunking
-> metadata enrichment
-> embeddings
-> vector database
-> retriever
-> consciousness-aware prompt builder
-> agent API
-> web interface
-> evaluations

Current Status

The repository has completed Phase 1 (foundation) and is actively in Phase 2 — Source Operations.

What exists today:

  • monorepo structure, editorial docs, consciousness core v1 (approved manifest), ontology starters
  • Markdown validation, tests, and CI
  • canon catalogs for 108 Upanishads, 18 Mahapuranas, 5 Vedic collections, Siddha corpus, Bhagavad Gita
  • large public-domain imports: Rig Veda Mandalas 1-4, principal Upanishads (Isha, Kena, Katha, Mundaka, Prashna, Mandukya, Taittiriya, Chandogya, Shvetashvatara, Brihadaranyaka, Kaivalya), Bhagavad Gita (18 chapters), Vishnu/Garuda Purana samples, Tirumandiram (payiram + tantras 1-9), Siddha corpus expansions
  • the Complete Works of Sri Aurobindo (CWSA, 37 volumes): full Markdown text for public-domain volumes, metadata records for volumes under active Trust copyright, via manifest-driven Ashram PDF ingestion
  • the Collected Works of the Mother (CWM, 18 volumes): metadata records with a local-only full-text pipeline, per docs/copyright_policy.md
  • automated ingestion scripts for Wikisource SBE, Tamil Wikisource and the Ashram PDF libraries
  • normalization, chunking, TF-IDF + dense retrieval (optional Qdrant), agent API scaffold with web UI
  • evaluation assets: golden questions, response-quality checks, and the Darshan response-contract harness (packages/evals)

See docs/content_import_roadmap.md and docs/source_import_status.md for the full import plan and status.

Site público

O website institucional fica em site/ e publica automaticamente no GitHub Pages via Actions (.github/workflows/deploy-pages.yml).

URL: https://sraphaz.github.io/aadi-yogi/

Uma vez: em Settings → Pages → Source, escolha GitHub Actions. Depois disso, cada push em main que altere site/** faz o deploy.

Validate Markdown

python scripts/validate/validate_markdown.py
pytest
ruff check .

Interface Vision

The application layer is designed as Darshan, an interface whose behavior embodies the wisdom of the corpus instead of talking about it: silence before speech, one question at a time, presence metrics instead of engagement metrics. See docs/darshan_interface_concept.md, docs/darshan_interface_spec.md, and the Sky-Forge session package in docs/skyforge/darshan/.

Consciousness foundation (for other repos)

Not a product checklist. A living basis of conduct from the texts.

# Python — install the foundation, consult a situation, send feedback
PYTHONPATH=. python -c "from packages.consciousness import load_foundation; print(load_foundation().agent_preamble[:400])"

# MCP — see .mcp.json and apps/mcp-server/README.md
PYTHONPATH=. python -m packages.consciousness.mcp_server

# HTTP
# GET  /consciousness/foundation
# POST /consciousness/consult
# POST /consciousness/feedback

Host repos declare .consciousness/link.yaml. Learning returns through reviewed feedback (content/consciousness_feedback/). ADR: spec/decisions/0006-consciousness-plugin-service.md.

Next Milestones

  • expand ontology and living maps
  • deepen retrieval and evaluation datasets
  • publish consciousness package for external installers
  • harden hosted API auth for cross-repo consumers

About

Source-grounded spiritual knowledge repository and AI consciousness architecture.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages