aurora-se 0.1.0

Autonomous Unified Repo Orchestrator for Reasoning & Adaptation — Self-Editing

AURORA-SE

Autonomous Unified Repo Orchestrator for Reasoning & Adaptation — Self-Editing.

Overview

AURORA-SE is a self-adapting engineering system that plans, applies, verifies, and learns from code changes across large-scale monorepos. It orchestrates a PDCA loop backed by observability, governance, and compliance controls.

Tech Stack Highlights

• Planner: DeepSeek-R1 via Ollama with optional GPT-4o-mini/Gemini critic stack
• Repo Intelligence: tree-sitter, pgvector, optional Milvus, Neo4j adapter
• Execution: Docker/Firecracker sandbox, CI profiles (fast/balanced/thorough)
• Observability: OpenTelemetry, Prometheus, Grafana dashboards
• Security: Syft, Grype, TruffleHog, license/policy engine
• Learning: Filtered-SFT with LoRA/PEFT adapters, federated sync
• Evaluation: SWE-Bench Lite/Live nightly/weekly suites

Getting Started

1. Install dependencies: pip install -e .[dev]
2. Install hooks: pre-commit install
3. Start services: docker compose up -d
4. Run aurora-se init to bootstrap configs
5. See docs/quickstart.md for workflow details

Packaging & Release

• Build artifacts: scripts/build_package.sh
• Generate checksums: python scripts/generate_checksums.py dist
• Preview docs locally: mkdocs serve
• See docs/launch/release_checklist.md for full launch steps.