Your risk, today.
- "What's my riskiest attack path?"get_attack_paths
- "Which critical Issues are due this week?"get_issues
- "Write the Terraform for Issue 412."get_fix
- "What can reach customer-data?"get_inventory
Ask about your cloud risk and your AI spend from Claude, Cursor, GitHub Copilot or Codex. I answer from your live Alye data, explain what it means, and draft the change for you to review.
No dashboards to open and no query language to learn. Ask the way you'd ask a colleague.
Your editor's assistant picks the right tool for your question. Every tool reads; none of them can change your cloud.
| Tool | Module | Returns |
|---|---|---|
get_attack_paths | Cloud security | Every route from the internet to sensitive data, with each route's weakest link |
get_issues | Cloud security | Issues ranked by reach, with severity, environment and due date |
get_fix | Cloud security | The fix for an Issue as CLI, Terraform, CloudFormation or console steps, filled in |
get_inventory | Cloud security | Resources, what faces the internet, which role they run as and what it can read |
get_scan_status | Cloud security | Progress and results of the latest scan for each cloud account |
get_scan_changes | Cloud security | What opened and closed since the previous scan |
get_llm_spend | LLM observability | Spend by service, feature, model or customer, over any period |
get_traces | LLM observability | Traces for a service or customer, with each call's cost and latency |
get_alerts | LLM observability | Alert rules and what fired, with the cause |
get_report | Both | A summary of open risk and AI spend for a week or month |
// .cursor/mcp.json
{
"mcpServers": {
"alye": {
"url": "https://api.alye.ai/mcp",
"headers": { "Authorization": "Bearer ${ALYE_API_KEY}" }
}
}
}# Claude Code
claude mcp add --transport http alye https://api.alye.ai/mcp \
--header "Authorization: Bearer $ALYE_API_KEY"// .vscode/mcp.json (GitHub Copilot agent mode)
{
"servers": {
"alye": {
"type": "http",
"url": "https://api.alye.ai/mcp",
"headers": { "Authorization": "Bearer ${env:ALYE_API_KEY}" }
}
}
}# Claude app
Settings › Connectors › Add custom connector
Name Alye
URL https://api.alye.ai/mcpAsk for the fix, get it as a change to your Terraform, review it and ship it in the same pull request.
"What changed overnight?" gives you new paths, closed ones and what's due, before your first meeting.
When finance asks why AI spend moved, ask me and forward the answer, numbers included.
Something else? Write to sales@alye.ai.
Any MCP client that supports remote servers over HTTP, including Claude, Claude Code, Cursor, GitHub Copilot in VS Code and Codex.
No. Every tool is read-only. When you ask for a fix, I draft it for you to review and apply in your own pipeline.
No. I receive the tool calls your assistant makes, such as "get the fix for Issue 412", and return your Alye data. Your editor's assistant writes the change in your files.
Answers come only from the organization your API key belongs to. Revoke the key in Alye and access stops.
No. Discover with Alye works with whichever modules you use. Tools for a module you don't use simply return nothing.
First trace right away. First ranked scan in about fifteen minutes.