What techniques does the skill use for investigation?

The skill applies structured methodologies including the 5 Whys, Fishbone diagrams, Bradford Hill criteria, and Directed Acyclic Graphs (DAGs).

Can I use this for software incident post-mortems?

Yes, it is designed to help trace system outages back to their fundamental source, distinguishing immediate triggers from the actual root cause.

How does this skill distinguish between correlation and causation?

It uses counterfactual reasoning ('What would have happened without X?') and identifies confounding variables that might cause two unrelated metrics to move together.

When is the best time to invoke this skill?

Trigger this skill whenever you are asking 'Why did this happen?'—especially when dealing with complex failures, unexpected data shifts, or recurring issues.

Causal Inference & Root Cause Analysis

Name: Causal Inference & Root Cause Analysis
Author: lyndonkl

bylyndonkl

•

分析与监控

Systematically investigates causal relationships to identify fundamental root causes and distinguish them from symptoms or correlations.

关于

This skill provides a rigorous framework for identifying the true drivers behind system failures, performance regressions, or shifts in data. By applying causal inference methodologies—such as counterfactual reasoning, Directed Acyclic Graphs (DAGs), and the Bradford Hill criteria—this skill helps users move beyond surface-level symptoms to find actionable root causes. It is particularly valuable for engineers investigating production incidents, data scientists analyzing metric shifts, and developers debugging complex, non-deterministic software behavior.

主要功能

Utilizes systematic frameworks like the 5 Whys, Fishbone diagrams, and Timeline analysis
8 GitHub stars
Distinguishes genuine causation from spurious correlations using counterfactual analysis
Generates comprehensive RCA documentation including evidence assessment and confidence levels
Identifies and controls for confounding variables and third-party factors
Builds structured causal models to map complex dependency chains (A -> B -> C)

使用场景

Analyzing shifts in business KPIs like conversion rates, user churn, or revenue drops
Debugging non-obvious software bugs and race conditions in high-concurrency environments
Investigating production outages or performance degradation in complex distributed systems

关于

主要功能

Utilizes systematic frameworks like the 5 Whys, Fishbone diagrams, and Timeline analysis
8 GitHub stars
Distinguishes genuine causation from spurious correlations using counterfactual analysis
Generates comprehensive RCA documentation including evidence assessment and confidence levels
Identifies and controls for confounding variables and third-party factors
Builds structured causal models to map complex dependency chains (A -> B -> C)

使用场景

Analyzing shifts in business KPIs like conversion rates, user churn, or revenue drops
Debugging non-obvious software bugs and race conditions in high-concurrency environments
Investigating production outages or performance degradation in complex distributed systems