Specora Core

Specora Core Documentation

Specora Core is an LLM-native Contract-Driven Development engine. Write YAML contracts, generate a complete production application. Contracts are the source of truth – code is a derived, disposable artifact.

The workflow: Install specora-core, scaffold a project with specora-init, open your LLM coding agent (Claude Code, Cursor, Windsurf) in the project directory, and talk to it. The LLM reads CLAUDE.md and operates everything via Python API. No CLI needed.


I want to…

Goal Read this
Create a new project from scratch Getting Started
Understand the 5-tier architecture Architecture
Write contracts (entity, workflow, route, page) CLAUDE.md – Contract Language Reference
See every contract field and option Contract Language Reference
Deploy to production with Docker Production Deployment
Understand the self-healing loop Self-Healing Loop
Set up webhook notifications Webhooks
Understand database migrations Migrations
Understand frontend generation Frontend Generation
Reverse-engineer an existing codebase Extractor
Set up an LLM provider LLM Providers
Use CLI commands (CI/CD, terminal) CLI Reference

How It Works

1. pip install specora-core
2. specora-init my_app
3. cd my_app
4. Open your LLM (Claude Code, Cursor, Windsurf)
5. Talk to the LLM -- it reads CLAUDE.md and does everything
6. docker compose up -d  -- boots the generated app

The LLM writes contracts, validates them, compiles to IR, generates production code, and manages the self-healing loop. All via Python function calls, no CLI required.


The Five Tiers

Tier Name What It Does LLM Required
1 Forge Compile contracts, generate code. Deterministic. No
2 Factory LLM-powered contract authoring from natural language. Yes
3 Healer Self-healing: detect errors, propose contract fixes, apply. Tier 1: No, Tier 2-3: Yes
4 Extractor Reverse-engineer existing code into contracts. No
5 Advisor Proactive evolution from telemetry. Planned

Documentation Index

Getting Started

Reference

Architecture

Production

Self-Healing

Code Generation

Reverse Engineering

LLM Operating Manual