Skip to content

ECOSYSTEM.md

Latest commit

 

History

History
58 lines (42 loc) · 2.26 KB

ECOSYSTEM.md

File metadata and controls

58 lines (42 loc) · 2.26 KB
Raw

DataCert AI Ecosystem

DataCert's local AI stack is split across three focused systems:

System Role Owns
AI Model DB Registry and directory Models, providers, capabilities, pricing, benchmarks, MCP servers, skills, install metadata
Context Engine Local continuity layer Handoffs, memory, rules, skills, modes, MCP tools, generated context, compiled file fallbacks
DRAM Runtime orchestration layer API daemons, local models, hosted model calls, process health, routing, logs, queues

The short version:

AI Model DB knows what exists. Context Engine keeps working context portable. DRAM runs and routes the systems that execute it.

Boundaries

Context Engine should stay focused on what host apps and agents can carry across sessions:

  • handoffs for active work
  • active skills
  • persistent memory
  • behaviour rules
  • mode/profile selection
  • runtime MCP tools for search, skill lookup, and status
  • generated manifests such as CONTEXT.md
  • compiled file fallbacks for Claude Code, Codex, Cursor, Windsurf, and other AI tools

AI Model DB should be the source of registry truth:

  • model metadata
  • provider metadata
  • local or hosted availability
  • pricing and capability fields
  • MCP server directory data
  • skill and workflow pack metadata

DRAM should own execution:

  • model/API daemon lifecycle
  • local runtime health
  • routing and fallback
  • job queues
  • logs and process status
  • GPU/API budget awareness

Integration Direction

Context Engine should consume registry data from AI Model DB and runtime status from DRAM, but it should not duplicate either system. It should keep context portable across host apps; it should not become the host app where the work happens.

The product flow is:

AI Model DB -> Context Engine -> Host apps / DRAM
     |              |                |
 registry       continuity      work + execution

This keeps the ecosystem modular: each repo can stand alone, but together they form a local AI operating layer.