How mirulhakim/cognitive-auditable-security-execution-case-opensource-ai-automation-workflow-orchestration-platform works
CASE is presented as a human-in-the-loop workflow platform: a job request is planned, paused for approval, processed into a document, then paused for delivery approval. In the available tree, the only runtime implementation is server.py; it uses placeholder fetched data and document content, with in-memory workflow checkpoints. The README describes broader ingress, dashboard, connector, and format-specific generator capabilities, but their source paths are absent here, so they are not represented as implemented components. Edges reflect server.py calls and graph transitions; broader API actors are documented there or in the README.
Main entry points: server.py LangGraph Workflow
Components 11
Job API
- Job Endpoints
server.py
Workflow Engine
- LangGraph Workflow
server.py - Intent Planner
server.py - Data Approval
server.py - Data Fetch Placeholder
server.py - Document Placeholder
server.py - Delivery Approval
server.py
AI and State
- Ollama Model
- Memory Checkpointer
server.py
Other components
- Gmail / Telegram
- Approval Dashboard
Connections 10
- Gmail / Telegram → Job Endpoints: submits job
server.py - Approval Dashboard → Job Endpoints: reviews jobs
server.py - Job Endpoints → LangGraph Workflow: starts or resumes
server.py - LangGraph Workflow → Intent Planner: dispatches
server.py - Intent Planner → Ollama Model: plans with
server.py - LangGraph Workflow → Data Approval: dispatches
server.py - LangGraph Workflow → Document Placeholder: dispatches
server.py - LangGraph Workflow → Delivery Approval: dispatches
server.py - Data Approval → Data Fetch Placeholder: returns sample data
server.py - LangGraph Workflow → Memory Checkpointer: persists state
server.py