How xdzy-td/aetheris-aiengine works
Aetheris is an agentic remote-sensing analysis service: a user submits a question and imagery, the controller plans and executes specialist tools, then returns an answer with evidence and confidence. A separate SAT-OS API adds answerability checks, evidence classification, falsification, verification, and certificates around the analysis pipeline. The repository also contains a modular hardening pipeline and several user interfaces. Edges reflect README statements and shown or indexed references; some module internals are unsampled, so uncertain calls are omitted.
Main entry points: planner.py Agentic Planner · sat_pipeline.py SAT-OS Orchestrator · app.py FastAPI · console.py Text Console
Components 24
User Interfaces
- FastAPI
interfaces/api/app.py - Text Console
interfaces/ui/console.py - Voice I/O
interfaces/voice - SAT-OS Routes
interfaces/api/sat_routes.py
Agentic Controller
- Agentic Planner
controller/planner.py - LLM Planner
controller/llm_planner.py - Ollama Client
controller/ollama_client.py - Policy Engine
controller/policy.py - Tool Executor
controller/executor.py - Tool Registry
controller/tool_registry.py - Routing Memory
controller/memory.py - Audit Trail
controller/audit_log.py - Artifact Store
controller/artifacts.py - Model Registry
controller/model_registry.py
Imagery Analysis
- Specialist Tools
specialists - Imagery I/O
specialists/_imagery.py - Image Preprocessing
preprocessing - Confidence Scoring
confidence
Pipeline Services
- SAT-OS Orchestrator
pipeline/services/sat_pipeline.py - Evidence & Verification
pipeline/modules - Aetheris Pipeline
pipeline/services/pipeline.py - Result Contracts
pipeline/schemas/contracts.py
Other components
- Analyst
- Ollama Service
Connections 25
- Analyst → FastAPI: submits query
README.md - Analyst → Text Console: submits query
README.md - FastAPI → Agentic Planner: routes requests
interfaces/api/app.py - Text Console → Agentic Planner: shares wiring
- Agentic Planner → LLM Planner: requests plans
controller/planner.py - LLM Planner → Ollama Client: requests completion
controller/llm_planner.py - Ollama Client → Ollama Service: sends requests
controller/ollama_client.py - Agentic Planner → Policy Engine: validates plan
controller/planner.py - Agentic Planner → Tool Executor: executes plan
controller/planner.py - Agentic Planner → Tool Registry: loads tools
controller/planner.py - Tool Executor → Specialist Tools: invokes tools
README.md - Specialist Tools → Imagery I/O: uses imagery I/O
specialists/_imagery.py - Agentic Planner → Confidence Scoring: scores results
controller/planner.py - Agentic Planner → Audit Trail: records execution
controller/planner.py - Agentic Planner → Routing Memory: uses routing memory
controller/planner.py - Agentic Planner → Artifact Store: stores artifacts
interfaces/api/app.py - Agentic Planner → Model Registry: resolves models
controller/planner.py - Analyst → SAT-OS Routes: submits SAT query
interfaces/api/sat_routes.py - SAT-OS Routes → SAT-OS Orchestrator: runs analysis
interfaces/api/sat_routes.py - SAT-OS Orchestrator → Aetheris Pipeline: runs inner pipeline
pipeline/services/sat_pipeline.py - SAT-OS Orchestrator → Evidence & Verification: classifies and verifies
pipeline/services/sat_pipeline.py - SAT-OS Orchestrator → Result Contracts: builds request
pipeline/services/sat_pipeline.py - Aetheris Pipeline → Result Contracts: returns module results
pipeline/services/pipeline.py - FastAPI → Image Preprocessing: supports preprocessing
- Voice I/O → FastAPI: provides voice input