Skip to main content
POST
/
v1
/
agents
curl -X POST https://api.runtools.ai/v1/runtimes \
  -H "X-API-Key: rt_live_xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "slug": "code-assistant",
    "name": "Code Assistant",
    "systemPrompt": "You are an expert software engineer...",
    "model": "claude-opus-4-5",
    "tools": ["bash", "read_file", "edit_file", "grep", "glob"]
  }'

{
  "data": {
    "id": "rt_abc123",
    "slug": "code-assistant",
    "name": "Code Assistant",
    "model": "claude-opus-4-5",
    "version": 1,
    "createdAt": "2025-01-21T10:00:00Z"
  }
}
The actual endpoint is POST /v1/agents (not /v1/runtimes). Agents are upserted by org + slug. The request body includes slug, name, model, systemPrompt, tools (array of tool names), sandboxId, and optional configJson with AI SDK parameters.

​
Request Body

​
slug
string
required
Unique identifier for the runtime (URL-safe)
​
name
string
required
Display name
​
systemPrompt
string
required
System prompt / instructions for the agent
​
model
string
required
Model to use (e.g., claude-opus-4-5, gpt-4o, gemini-2.0-flash)
​
tools
array
required
Array of tool names to enable
​
modelConfig
object
Model-specific configuration

​
Response

​
id
string
Runtime ID
​
slug
string
Runtime slug
​
version
number
Current version number
​
createdAt
string
ISO 8601 timestamp
curl -X POST https://api.runtools.ai/v1/runtimes \
  -H "X-API-Key: rt_live_xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "slug": "code-assistant",
    "name": "Code Assistant",
    "systemPrompt": "You are an expert software engineer...",
    "model": "claude-opus-4-5",
    "tools": ["bash", "read_file", "edit_file", "grep", "glob"]
  }'

{
  "data": {
    "id": "rt_abc123",
    "slug": "code-assistant",
    "name": "Code Assistant",
    "model": "claude-opus-4-5",
    "version": 1,
    "createdAt": "2025-01-21T10:00:00Z"
  }
}