Links:
- PyPI: https://pypi.org/project/reskpoints
- GitHub: https://github.com/Resk-Security/ReskPoints
- RESK Security: https://resk.fr
AI agents are getting more autonomous every week. But with autonomy comes a visibility problem: what did your agent actually do?
ReskPoints is an open source Python library that traces every agent action with sampling, masking, and multi-export. It gives you observability without the overhead.
Quick Start
pip install reskpoints
Then decorate your agent functions:
from reskpoints import trace, configure
configure(exporters=["console", "datadog"])
@trace(sample_rate=0.5)
def search_knowledge_base(query: str):
# Your agent logic here
return results
Key Features
- Adaptive Sampling — control verbosity per function so hot paths dont flood your logs
- Built-in Masking — redact API keys, PII, or any pattern before data leaves your process
- Multi-Export — Console, Datadog, Prometheus, OpenTelemetry, file, and webhooks all supported from one config
- YAML Config — no code changes to switch logging backends
# config.yaml
exporters:
- console
- datadog:
api_key: ${DD_API_KEY}
service: my-agent
sampling:
default_rate: 0.1
overrides:
search_knowledge_base: 1.0
masking:
patterns:
- "sk-[A-Za-z0-9]+"
- "\\b\\d{16}\\b"
Why Agent Logging Matters
When an autonomous agent hallucinates a tool call or leaks data through a prompt injection, you need the trace to understand what happened. ReskPoints gives you that trace with minimal overhead.
Install it today and start seeing what your agents actually do.
pip install reskpoints
Check the GitHub repo for full docs and examples.
Top comments (0)