# MCP + A2A — Workspace Initiative

This folder contains a workspace-wide scan of the ByteLyst ecosystem (platform + shared packages + dashboards + product repos) for opportunities to leverage:

- MCP (Model Context Protocol)
- A2A (agent-to-agent) patterns

## Documents

- `WORKSPACE_USE_CASE_CATALOG.md`
- `MCP_SERVER_FRAMEWORK.md`
- `A2A_ORCHESTRATION_FRAMEWORK.md`
- `IMPLEMENTATION_PLAN.md`
- `DOMAIN_PLATFORM_SERVICE.md`
- `DOMAIN_EXTRACTION_SERVICE.md`
- `DOMAIN_DASHBOARDS.md`
- `DOMAIN_PACKAGES_AND_SDKS.md`
- `DOMAIN_PRODUCTS.md`

## Scope of scan

Primary sources used:

- `services/platform-service` (telemetry, diagnostics, jobs, settings, webhooks, auth, etc.)
- `services/extraction-service` (sidecar + tasks + async jobs + rate limits)
- `packages/*` (Swift/Kotlin platform SDKs, TS clients, event bus, telemetry/diagnostics clients)
- `dashboards/admin-web`, `dashboards/tracker-web`, `dashboards/ux-lab`
- Product repos: ChronoMind (`learning_ai_clock`), NomGap (`learning_ai_fastgap`), PeakPulse (`learning_ai_peakpulse`), LysnrAI (`learning_voice_ai_agent`), MindLyst (`learning_multimodal_memory_agents`), JarvisJr (`learning_ai_jarvis_jr`)

## Notation

- **Tool**: an MCP tool callable by an LLM/agent.
- **Resource**: an MCP resource (read-only or read-mostly) exposed for context.
- **Prompt**: an MCP prompt template.
- **Agent**: an A2A-capable worker with a specific responsibility.