Track Every AI Agent Action with ReskPoints Python Logger

#ai #llm #python #opensource

Links:

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

The Problem

AI agents make dozens of tool calls in a single session. Debugging them with print statements and console.log is unsustainable. You need structured observability per-action timelines export pipelines and sensitive data masking.

The Solution: ReskPoints

ReskPoints is a drop-in Python decorator that wraps your agent functions with automatic tracing sampling and multi-export.

from reskpoints import trace

@trace(sample_rate=0.5, mask_fields=["api_key", "password"])
def search_knowledge_base(query: str) -> str:
    # Your agent logic here
    return results

@trace(export="datadog")
def call_llm(prompt: str) -> str:
    # LLM call
    return response

Key Features

Automatic sampling - Configure sampling rate per function to control volume
Sensitive data masking - Regex-based field masking before export
Multi-export - Datadog Prometheus OpenTelemetry console file and webhooks
Decorator-based - Zero boilerplate wraps any async or sync function
Context propagation - Track parent-child relationships across agent calls

Install

pip install reskpoints

Why It Matters

70% of organizations lack AI governance. Structured logging is the first step toward understanding what your agents actually do in production.

Check out the GitHub repo for full documentation and examples: https://github.com/Resk-Security/ReskPoints

DEV Community