MCP Server

Connect AI editors via the Model Context Protocol

Overview

The Sysevo MCP server exposes a Streamable HTTP endpoint that any MCP-compatible AI client can connect to. Once connected, your editor can read, create, and modify voice agents using natural language.

Endpoint

POST
https://api.sysevo.io/api/v1/mcp/
  • Transport: Streamable HTTP (MCP 2024-11 spec)
  • Auth: X-API-Key header

Client Configurations

Claude Code — ~/.claude/settings.json

json
{
  "mcpServers": {
    "sysevo-voice": {
      "type": "http",
      "url": "https://api.sysevo.io/api/v1/mcp/",
      "headers": { "X-API-Key": "YOUR_API_KEY" }
    }
  }
}

Cursor — ~/.cursor/mcp.json

json
{
  "mcpServers": {
    "sysevo-voice": {
      "url": "https://api.sysevo.io/api/v1/mcp/",
      "headers": { "X-API-Key": "YOUR_API_KEY" }
    }
  }
}

Claude Desktop — claude_desktop_config.json

Same format as Claude Code. Requires Claude Desktop 0.10+.

Zed — ~/.config/zed/settings.json

json
{
  "context_servers": {
    "sysevo-voice": {
      "url": "https://api.sysevo.io/api/v1/mcp/",
      "headers": { "X-API-Key": "YOUR_API_KEY" }
    }
  }
}

Test Connection

bash
curl -X POST https://api.sysevo.io/api/v1/mcp/ \
  -H "Content-Type: application/json" \
  -H "X-API-Key: YOUR_API_KEY" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}'

All 14 MCP Tools

Agent Tools

MethodPathDescription
TOOLlist_workflowsList all voice agents — active, archived, or all
TOOLget_workflowFetch an agent projected into structured view
TOOLget_workflow_codeGet the editable TypeScript SDK source for an agent
TOOLcreate_workflowParse TypeScript and create a new published agent
TOOLsave_workflowParse TypeScript and save as a new draft (published untouched)

Catalog Tools (Read-Only)

MethodPathDescription
TOOLlist_toolsList all HTTP/MCP tool integrations
TOOLlist_documentsList all knowledge base documents
TOOLlist_credentialsList credentials by name (no secret values)
TOOLlist_recordingsList call recordings, optionally by agent

Node Type Tools

MethodPathDescription
TOOLlist_node_typesList all pipeline node types with field specs
TOOLget_node_typeGet full field spec for a specific node type

Documentation Tools

MethodPathDescription
TOOLlist_docsBrowse documentation tree (top-level or by path)
TOOLread_docRead a documentation page by path
TOOLsearch_docsFull-text search across all documentation

Recommended Call Order

Editing an existing agent

  1. list_workflows — locate the agent
  2. get_workflow_code — fetch current TypeScript source
  3. (optional) list_node_types / get_node_type — check node field specs
  4. Mutate the source in the LLM's context
  5. save_workflow — persist as a new draft

Creating a new agent

  1. list_node_types — discover available node types
  2. get_node_type — check field specs for nodes you'll use
  3. Author TypeScript from scratch
  4. create_workflow — publishes immediately, returns workflow_id

When uncertain about how something works

  1. search_docs — keyword lookup (call this first)
  2. read_doc — fetch the full page
  3. list_docs — browse sections when queries are too broad
Was this page helpful?
Open Dashboard →
PreviousPublic DownloadsNextWorkflow SDK