Track Every AI Agent Action with ReskPoints - Agent Observability Done Right

Links:

• PyPI: https://pypi.org/project/reskpoints
• GitHub: https://github.com/Resk-Security/ReskPoints
• resk.fr: https://resk.fr

---

You deploy an AI agent. It calls tools. It hits APIs. It makes decisions. But what actually happened? Without structured logging you'll never know.

ReskPoints is a Python library that logs every agent action with configurable sampling, field masking, and multi-export. Think structured logging purpose-built for AI agents.

Quick Start

from reskpoints import AgentLogger

logger = AgentLogger(
    name="my-agent",
    sample_rate=0.5,       # Log 50% of actions
    mask_fields=["api_key", "user_email"],
    exporters=["console", "datadog"]
)

@logger.track("tool_call")
def search_database(query: str):
    # Your agent logic here
    return results

logger.info("Agent started", metadata={"session": "abc123"})

Key Features

• Configurable sampling - log 100% of actions in dev, 10% in prod
• Field masking - automatically redact PII, API keys, tokens from logs
• Multi-export - Datadog, Prometheus, OpenTelemetry, console, webhooks
• Decorator API - one decorator logs any function call
• Lightweight - pure Python, no heavy dependencies

pip install reskpoints

Agent observability is becoming mandatory as AI systems handle more production workloads. ReskPoints gives you the visibility you need without the complexity.

Check it out: PyPI | GitHub | resk.fr

What logging strategy do you use for your agents?