Skip to main content
Glama
caresolace

ai-care-agent-mcp

by caresolace
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Hybrid
README.md
# ai-care-agent-mcp

Headless care coordination agent: **MCP tools** and **A2A skills** for case-aware, multi-pathway provider search, powered by [ai-carematch](https://github.com/caresolace/ai-carematch). Case-management API integration is planned for later phases.

## Documentation

- **[Spec](docs/spec.md)** — architecture, tools, case model, phases
- **[Phased plan](docs/plan/phased-delivery.md)** — status, workstreams, exit criteria
- **[AWS deployment / IaC](docs/architecture/aws-deployment.md)** — Terraform, ECS, API Gateway
- **[Acceptance](docs/acceptance.md)** — program and phase acceptance criteria
- **[Project structure](docs/architecture/project-structure.md)** — repo layout and layer conventions

## Quick start

```bash
uv sync --dev --group test
cp .env.example .env   # set CAREMATCH_BASE_URL
make verify
make run-mcp           # stdio MCP server
```

## Layout

```text
src/ai_care_agent_mcp/
├── mcp/          # FastMCP tools + registry
├── carematch/    # upstream HTTP adapter
├── case/         # case profile resolution (Phase 2)
├── pathways/     # pathway catalog
├── core/         # settings, errors, logging
└── a2a/          # Agent Card (Phase 3)
```

## Status

| Phase | Name | Status |
|-------|------|--------|
| 0 | Foundation | Complete |
| 1 | Carematch MCP MVP | **In progress** — schemas, OpenAPI pin, validation |
| 2 | Case-backed search | Planned |
| 3 | A2A + HTTP + **AWS IaC** | Planned — ECS, API Gateway, Terraform |
| 4 | Case-management APIs | Planned |

Details: [`docs/plan/phased-delivery.md`](docs/plan/phased-delivery.md).
Deploy Server
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

B3.4/5.0

Scored across 4 tools

Disambiguation5/5

Each tool has a distinct and well-defined purpose: resolving case references, checking health, listing pathways, and searching providers. No two tools overlap in functionality.

Naming Consistency4/5

All tool names use snake_case and are mostly verb_noun (get_, list_, search_), with the exception of health_check which is a noun phrase, but still follows the same style.

Tool Count4/5

Four tools is a reasonable count for a focused care agent server. It covers the essential operations without being sparse or excessive.

Completeness4/5

The tool set covers the core workflow of obtaining search context, listing pathways, and searching providers. A health check is included. Minor gaps exist (e.g., no direct CRUD for cases), but the surface is adequate for its stated purpose.

Maintenance

ActivitySlowing
ResponsivenessSyncing

Resources