wirelog 0.3.0

WireLog analytics client — zero dependencies

wirelog

WireLog analytics client for Python. Zero dependencies — stdlib only.

Headless alternative to PostHog, Amplitude, and Mixpanel — designed for agents instead of dashboards.

Install

pip install wirelog

Quick Start

from wirelog import WireLog

wl = WireLog(api_key="sk_your_secret_key")

# Track an event (non-blocking, batched automatically)
wl.track("signup", user_id="u_123", event_properties={"plan": "free"})

# Query analytics (returns Markdown by default)
result = wl.query("signup | last 7d | count by day")
print(result)

# Identify a user (bind device → user, set profile)
wl.identify("[e-mail]", device_id="dev_abc", user_properties={"plan": "pro"})

# Flush remaining events on shutdown
wl.close()

Design Principles

This client is designed to never break your application:

• Non-blocking by default: track() buffers events and returns immediately
• Automatic batching: Events are sent in batches (default 10 per batch, or every 2 seconds)
• Bounded memory: Queue capped at 10,000 events — oldest events are dropped when full
• Retry with backoff: Transient failures (429, 5xx) are retried up to 3 times
• Graceful shutdown: close() flushes remaining events; also works as a context manager
• Background thread: Flush worker is a daemon thread — won't block process exit

Context Manager

with WireLog(api_key="sk_...") as wl:
    wl.track("signup", user_id="u_123")
# Events are flushed automatically on exit

Configuration

wl = WireLog(
    api_key="sk_...",              # Falls back to WIRELOG_API_KEY env var
    host="https://api.wirelog.ai", # Falls back to WIRELOG_HOST env var
    timeout=30,                    # HTTP timeout in seconds
    flush_interval=2.0,            # Seconds between auto-flushes (0 = sync mode)
    batch_size=10,                 # Max events per batch
    queue_size=10000,              # Max buffered events
    on_error=lambda e: print(e),   # Background error callback
    disabled=False,                # True = track() is a no-op
)

Synchronous Mode

Set flush_interval=0 to send each track() call immediately (blocking):

wl = WireLog(api_key="sk_...", flush_interval=0)
result = wl.track("signup", user_id="u_123")  # blocks, returns {"accepted": 1}

API

wl.track(event_type, *, user_id, device_id, session_id, event_properties, user_properties, insert_id, origin, client_originated)

Track a single event. Auto-generates insert_id and time if not provided.

wl.track_batch(events, *, origin=None, client_originated=None)

Track multiple events in one request (up to 2000). Always sends immediately.

wl.query(q, *, format="llm", limit=100, offset=0)

Run a pipe DSL query. Format: "llm" (Markdown), "json", or "csv".

wl.identify(user_id, *, device_id, user_properties, user_property_ops)

Bind a device to a user and/or update profile properties.

wl.flush()

Flush all buffered events. Blocks until the queue is drained.

wl.close()

Flush remaining events and stop the background thread. Idempotent.

Zero Dependencies

This library uses only the Python standard library (urllib.request, json, threading, queue, time, uuid, os). No requests, no httpx, no urllib3. It works out of the box on any Python 3.9+ installation.

Learn More

• WireLog — headless analytics for agents and LLMs
• Why WireLog vs PostHog/Amplitude — comparison guide
• Query language docs
• API reference