Diagnosing Experiment Results
Diagnoses bias, anomalies, and strange-looking results on a specific PostHog experiment. Covers empty / 0-exposure experiments, sample ratio mismatch, identity fragmentation, multi-variant exposure, uneven-split…
Overview
MD Skills lists Diagnosing Experiment Results as an agent skill in Data & Analytics. The source repository is PostHog/skills. The source listing includes a SKILL.md file. Open the original instructions to review what the skill asks the agent to do before installing it.
Getting started
Open the source file and follow the project’s installation instructions for your agent.
Skill instructions
Sources and documentation
- https://www.mdskills.ai/skills/diagnosing-experiment-results
- https://github.com/PostHog/skills/tree/main/skills/omnibus/diagnosing-experiment-results
Related MCP servers and skills
- Analyzing Experiment Session ReplaysAgent skill
- Configuring Experiment AnalyticsAgent skill
- Configuring Experiment RolloutAgent skill
- Nullary - Negative Results Intelligence For Drug DiscoveryMCP server
- Altss Public Profiles MCPMCP server
- Canopy API - Amazon Data for AgentsMCP server