How hypnofox/supportpublic- works
Supportnetes is an early-stage design for an AI orchestration layer that works across existing technical-support systems rather than replacing them. Its documented workflow connects source data, assembles case context and evidence-backed recommendations, applies policy and human approval, then writes approved actions back and captures outcomes for audit and evaluation. The repository contains documentation only, so this graph represents documented architecture, not verified runtime wiring; optional Redis is shown separately.
Main entry points: architecture.md Connectors · storage-and-retrieval.md Knowledge retrieval · product-principles.md Human approval
Components 18
Integration
- Connectors
docs/architecture.md - Ingestion
docs/architecture.md - Action writeback
docs/architecture.md
Case intelligence
- Parsing and classification
docs/architecture.md - Knowledge retrieval
docs/storage-and-retrieval.md - Case context
docs/architecture.md - Evidence-backed recommendations
docs/architecture.md - Workflow coordination
docs/architecture.md - Outcome capture
docs/architecture.md
Governance
- Identity and permissions
docs/architecture.md - Policy checks
docs/product-principles.md - Human approval
docs/product-principles.md - Audit and evaluation
docs/architecture.md
Data platform
- Relational storage
docs/storage-and-retrieval.md - Attachment storage
docs/architecture.md - Redis (later)
docs/architecture.md
Other components
- Support operator
- Support systems
Connections 20
- Support systems → Connectors: provide source data
- Connectors → Ingestion: deliver data
- Ingestion → Identity and permissions: supply records
- Identity and permissions → Parsing and classification: apply access context
- Parsing and classification → Relational storage: store structured data
- Parsing and classification → Attachment storage: store attachments
- Relational storage → Knowledge retrieval: retrieve records
- Attachment storage → Knowledge retrieval: retrieve attachments
- Knowledge retrieval → Case context: provide evidence
- Case context → Evidence-backed recommendations: inform next steps
- Evidence-backed recommendations → Workflow coordination: inform workflow
- Workflow coordination → Policy checks: check actions
- Policy checks → Human approval: request approval
- Support operator → Human approval: review actions
- Human approval → Action writeback: authorize actions
- Action writeback → Connectors: dispatch actions
- Connectors → Support systems: write approved actions
- Support systems → Outcome capture: return outcomes
- Outcome capture → Audit and evaluation: inform evaluation
- Redis (later) → Workflow coordination: support transient state