Tekhnika/agentic-ai-observability-free

---
pretty_name: Agentic AI Observability
language:
- en
license: cc-by-nc-4.0
tags:
- tabular-data
- analytics
- observability
- agentic-ai
task_categories:
- tabular-classification
- time-series-forecasting
size_categories:
- 1K<n<10K
---

# Agentic AI Observability

Free sample for AI agent reliability dashboards, anomaly exploration, and observability-oriented analytics workflows.

## What is included
- agents.csv: 46 rows, 8 columns
- events.csv: 5590 rows, 8 columns
- runs.csv: 1863 rows, 8 columns

## Why this dataset is useful
- Good starter sample for an agent reliability dashboard or observability notebook.
- Useful for validating run-level and event-level analytics in Python, SQL, and BI tools.
- Lightweight enough for quick experiments while still matching the core workflow of the full starter pack.

## Starter use cases
- Agent reliability baseline using run and event data.
- Observability dashboard for event severity, run behavior, and agent health patterns.

## Schema overview
### agents.csv
- Rows: 46
- Columns: agent_id, agent_name, agent_type, deployment_env, region, status, traffic_tier, reliability_profile

### events.csv
- Rows: 5590
- Columns: event_id, run_id, agent_id, event_time, event_type, severity, component_name, queue_depth

### runs.csv
- Rows: 1863
- Columns: run_id, agent_id, started_at, duration_ms, run_status, failure_risk_score, slo_breach_flag, anomaly_window_flag

## Free vs full version
- Free Kaggle sample: reduced rows, reduced columns, starter notebook, and enough linked observability tables to validate the core workflow.
- Full version: full row volume, richer feature coverage, tool and feedback tables, and extra starter assets for dashboard, SQL, and anomaly-analysis work.

## Upgrade to full version
- Full version: https://tekhnikalab.gumroad.com/l/agentic-ai-observability
- Upgrade if you need the full linked schema plus starter assets that get you to a dashboard, SQL project, or anomaly baseline faster.

## Notes
- Contains generated data only and no real personal data.
- Designed as a lightweight free sample for evaluation and discovery.