How munch2u-a11y/aimaos works
AIMAOS is a local-first office workstation: a user submits matter requests or files through the browser UI, work is recorded and routed to specialist agents, and resulting drafts or updates return for human review. The map includes the documented matter, scheduling, document, security, and memory capabilities. Some subsystems are represented by their exact source paths but were not sampled; their links are included only where the README or SOURCE INDEX supports them.
Main entry points: aimaos_ui.py Workstation API · case_agent.py Matter reviewer · belief_store.py Belief store · orchestrator.py Marley scheduler
Components 26
Workstation
- Web UI entry
ui/web_ui.py - Workstation API
aimaos_ui.py - Document review
core/document_review.py - Agent status views
core/agent_widgets.py
Work Coordination
- Background jobs
core/jobs.py - Office Board
core/comms/office_board.py - Daily workflow review
core/workflow_review.py - Marley scheduler
starter_packs/document_heavy/Marley-AI/core/orchestrator.py
Specialist Work
- Shared agent kernel
core/office_agent.py - Tool delegation
core/delegation.py - Matter reviewer
core/case_agent.py - Drive ingestion
starter_packs/document_heavy/Kai-AI/tools/drive_ingestion.py - Matter records
starter_packs/document_heavy/Kai-AI/business/client_file.py - Document production
starter_packs/document_heavy/Alix-AI/business/document_engine.py - Template review
starter_packs/document_heavy/Alix-AI/business/subagents/template_reviewer.py
Agent Memory
- Memory ingestion
core/mrag/core/memory_ingestor.py - Belief store
core/mrag/memory/belief_store.py - Belief consolidation
core/mrag/core/belief_consolidator.py - Memory retrieval
core/mrag/core/pre_generative_injection.py - Vector retrieval
core/mrag/core/vector_store.py - Token counting
core/mrag/core/token_counting.py
Local Platform
- Safe access policy
core/security.py - Office database
core/db/office_sqlite.py - Local model access
core/llm.py - Safe local writes
core/atomic_io.py
Other components
- Office user
Connections 30
- Office user → Web UI entry: uses workstation
README.md - Web UI entry → Workstation API: forwards requests
ui/web_ui.py - Workstation API → Background jobs: queues model work
aimaos_ui.py - Workstation API → Office Board: uses task board
aimaos_ui.py - Workstation API → Office database: reads and writes
aimaos_ui.py - Workstation API → Document review: serves review
aimaos_ui.py - Workstation API → Safe access policy: checks access
aimaos_ui.py - Workstation API → Agent status views: renders status
aimaos_ui.py - Workstation API → Daily workflow review: uses review logic
aimaos_ui.py - Office Board → Office database: syncs tasks
core/comms/office_board.py - Marley scheduler → Office Board: reads tasks
starter_packs/document_heavy/Marley-AI/core/orchestrator.py - Marley scheduler → Daily workflow review: runs daily review
- Marley scheduler → Shared agent kernel: dispatches agents
- Shared agent kernel → Tool delegation: delegates tools
- Tool delegation → Local model access: plans tool work
- Drive ingestion → Matter records: updates records
starter_packs/document_heavy/Kai-AI/tools/drive_ingestion.py - Drive ingestion → Matter reviewer: requests matter review
starter_packs/document_heavy/Kai-AI/tools/drive_ingestion.py - Matter reviewer → Shared agent kernel: uses shared kernel
core/case_agent.py - Matter reviewer → Belief store: reads and writes memory
core/case_agent.py - Matter reviewer → Memory retrieval: retrieves context
core/case_agent.py - Matter reviewer → Vector retrieval: uses retrieval index
core/case_agent.py - Document production → Template review: supports template review
- Memory ingestion → Belief store: stores event chunks
core/mrag/core/memory_ingestor.py - Memory ingestion → Token counting: counts tokens
core/mrag/core/memory_ingestor.py - Belief consolidation → Belief store: consolidates beliefs
core/mrag/core/belief_consolidator.py - Belief consolidation → Memory retrieval: syncs retrieval context
core/mrag/core/belief_consolidator.py - Belief consolidation → Token counting: counts tokens
core/mrag/core/belief_consolidator.py - Belief store → Vector retrieval: indexes for retrieval
- Belief store → Safe local writes: persists memory
- Matter records → Office database: upserts case data
starter_packs/document_heavy/Kai-AI/business/client_file.py