Skip to main content

Flow Insights

Flow insights define the smallest shared language for understanding how work moves from proposal to production across providers. The goal is a provider-neutral model and signal catalog that let adapters, inference logic, and UIs converge without locking into any one platform shape.

What this covers​

  • Canonical concepts that change action: work item, change, actor, review state, validation state, evidence state, ownership state, and risk signal.
  • The initial signal catalog for the MVP, with semantic inference rules and worked examples.
  • Scope guardrails that prevent CMDB drift and workflow-engine drift.
  • Layer 1 intent scenarios that express the flow insights as Gherkin ground truth.
  • MVP boundaries, provider strategy, implementation strategy, and deferred scope.

Where to start​

Understanding the semantic foundation​

  • Read the canonical model to understand the shared vocabulary.
  • Review the signal catalog to see how the model is applied.
  • Read the adapter contract to understand the normalized types that provider adapters must produce.
  • Use the scope guardrails to keep provider-specific or workflow-shaped concerns out of the core.
  • Explore the intent scenarios to see the Gherkin statements that anchor the MVP.
  • Reference the HTTP API contract for the list and detail endpoints that expose flow insights to clients.

Understanding the MVP boundaries and strategy​

Architecture decisions​

Epic alignment​

The flow insight MVP depends on this language before any adapter, inference engine, or UI work begins. This section is semantic only — no storage schemas, HTTP routes, MCP tools, or UI behaviors are defined here.