plott-sdk-core 0.0.1a1

Plott Analytics SDK Core for Python - Event tracking and batching

Plott SDK Core for Python

The core analytics client for Plott, providing event tracking with batching, retry logic, and async support.

Installation

pip install plott-sdk-core

Quick Start

from plott_analytics import PlottAnalytics

# Initialize the client
client = PlottAnalytics({
    "api_key": "cpk_...",  # or set PLOTT_API_KEY env var
    "environment": "development",
})

# Track events
client.track_message_event(
    role="user",
    content="Hello, world!",
    run_id="run-123",
    session_id="session-456",
)

client.track_message_event(
    role="assistant",
    content="Hi there! How can I help you today?",
    run_id="run-123",
    session_id="session-456",
)

# Flush and shutdown when done
await client.flush()
await client.shutdown()

Event Types

Message Events

Track messages between users and AI assistants:

client.track_message_event(
    role="user",  # or "assistant", "system"
    content="Message content",
    run_id="run-123",
    session_id="session-456",
    token_count=50,  # optional
    model="gpt-4",   # optional
)

Run Events

Track the lifecycle of agent runs:

client.track_run_event(value="start", run_id="run-123")
# ... agent execution ...
client.track_run_event(value="end", run_id="run-123")

Tool Events

Track tool/function calls:

client.track_tool_event(
    tool_call_id="call-123",
    tool_status="started",
    tool_name="search",
    args={"query": "latest news"},
)

client.track_tool_event(
    tool_call_id="call-123",
    tool_status="completed",
    result="Found 10 results...",
)

State Snapshot Events

Track state changes:

client.track_state_snapshot_event(
    snapshot={"messages": [...], "context": {...}},
    run_id="run-123",
)

Error Events

Track errors:

client.track_error_event(
    error={
        "name": "ValidationError",
        "message": "Invalid input",
        "code": "INVALID_INPUT",
    },
    run_id="run-123",
    severity="medium",
)

LLM Call Events

Track LLM API calls with token usage:

client.track_llm_call_event(
    model="gpt-4",
    input_tokens=100,
    output_tokens=50,
    latency_ms=1200,
    provider="openai",
    is_streaming=True,
)

Configuration

client = PlottAnalytics({
    "api_key": "cpk_...",           # Required
    "environment": "production",     # production, staging, development, test
    "retry_attempts": 3,             # Max retry attempts
    "retry_delay": 1.0,              # Initial retry delay (exponential backoff)
    "debug": False,                  # Enable debug logging
})

License

MIT