Agent/Session Management

Cylestio Perimeter supports both automatic and manual management of Agent IDs and Session IDs. By default, it automatically generates unique Agent and Session IDs for each conversation. However, you can override this behavior at any time by explicitly specifying your own Agent ID or Session ID via HTTP headers.

Key Features

• Automatic Session Detection: Sessions are automatically detected based on message history and conversation patterns
• Automatic Agent Identification: Agents are automatically identified using prompt hashing
• Manual Overrides: Custom session and agent IDs can be provided via HTTP headers
• Consistent Tracking: All events maintain consistent session and agent identifiers throughout the conversation

Overriding Agent and Session IDs

Override automatic behavior by providing custom IDs via HTTP headers:

Available Headers

Header	Purpose	Example Value
x-cylestio-session-id	Set custom session identifier	user-123-conversation-1
x-cylestio-agent-id	Set custom agent identifier	math-tutor-v2

Usage Examples

Custom Agent ID Only:

curl -X POST "http://localhost:3000/v1/messages" \
  -H "Content-Type: application/json" \
  -H "anthropic-version: 2023-06-01" \
  -H "x-cylestio-agent-id: math-tutor-v2" \
  -d '{
    "model": "claude-3-sonnet-20240229",
    "system": "You are a math tutor.",
    "messages": [
      {"role": "user", "content": "What is 2+2?"}
    ],
    "max_tokens": 100
  }'

Custom Both IDs:

curl -X POST "http://localhost:3000/v1/messages" \
  -H "Content-Type: application/json" \
  -H "anthropic-version: 2023-06-01" \
  -H "x-cylestio-session-id: user-123-lesson-1" \
  -H "x-cylestio-agent-id: math-tutor-v2" \
  -d '{
    "model": "claude-3-sonnet-20240229",
    "system": "You are a math tutor.",
    "messages": [
      {"role": "user", "content": "What is 2+2?"}
    ],
    "max_tokens": 100
  }'

Using APIs

Anthropic SDK

from anthropic import Anthropic

client = Anthropic(
    base_url=base_url,
    default_headers={
        "x-cylestio-agent-id": "math-agent",
        "x-cylestio-session-id": "math-agent-session-2",
    },
)

message = client.messages.create(
    model="claude-3-sonnet-20240229",
    max_tokens=1024,
    system="You are a math tutor.",
    messages=[{"role": "user", "content": "Hello, Claude!"}],
)

Note: OpenAI SDK example is coming soon.