SLA.directory MCP Server
# SLA.directory MCP server
An [MCP](https://modelcontextprotocol.io) server that lets AI assistants query **[SLA.directory](https://sla.directory)** in context: uptime commitments, service credits, claim windows, and **per-service** SLAs for cloud and SaaS vendors.
It reads the public SLA.directory API at runtime (with a short cache). No API key, no data bundled.
## Tools
- `search_vendors(query)`: find vendors and services by name or category.
- `get_vendor(slug)`: full SLA record for one vendor.
- `get_service_sla(slug, service?)`: per-service SLAs (e.g. AWS S3 vs EC2), optionally filtered.
- `compare_vendors(slugs[])`: key terms side by side (uptime, max credit, claim window, auto-credit).
- `list_categories()` and `vendors_in_category(category)`.
Vendor lookups are **fuzzy** (e.g. "Amazon Web Services" resolves to `aws`), answers are **concise and cited** (every result carries the official source URL and verification date), and tools are marked read-only so assistants can call them freely.
## Prompts
Ready-made flows you can pick in the client:
- `sla_review(vendor)`: summarize one vendor's SLA (uptime, credits, claim process, exclusions, weakest services).
- `compare_for_renewal(vendors)`: compare several vendors for a renewal decision.
- `weakest_services(vendor)`: find a vendor's lowest-uptime services for error-budget planning.
## Use it
Run with Node 18+ (no install needed once published):
```bash
npx sla-directory-mcp
```
### Claude Desktop / Cursor
Add to your MCP config:
```json
{
"mcpServers": {
"sla-directory": {
"command": "npx",
"args": ["-y", "sla-directory-mcp"]
}
}
}
```
Then ask things like *"What's AWS S3's SLA?"*, *"Compare Stripe, Adyen, and PayPal SLA credits"*, or *"Which security vendors offer automatic credits?"*
## Run from source
```bash
npm install
npm run build
node dist/index.js # stdio server
```
Point it at a different API base with `SLA_API_BASE` (defaults to `https://sla.directory/api/v1`).
## License
MIT. SLA data served by the API is licensed CC-BY-4.0 by SLA.directory. Always confirm against the vendor's official SLA before relying on this data for contractual or financial decisions.
TDQS
Scored across 6 tools
Each tool has a clear, distinct purpose: comparing vendors, fetching per-service SLAs, retrieving vendor details, listing categories, searching, and listing vendors by category. There is no overlap in functionality.
Most tools use a verb_noun pattern (compare_vendors, get_service_sla, get_vendor, list_categories, search_vendors). However, 'vendors_in_category' uses a noun_preposition_noun structure, breaking the pattern slightly.
With 6 tools, the set is well-scoped for an SLA directory. It covers essential operations without being too numerous or too few, providing a focused interface.
The tools cover searching, browsing by category, retrieving vendor details, per-service SLAs, and cross-vendor comparison. This represents a comprehensive read-only surface for the domain, with no obvious gaps.