The Agent Skills Directory

[SAFE]: A detailed analysis of the skill's instructions, code snippets, and metadata found no evidence of malicious behavior, prompt injections, or obfuscation. The content is strictly educational and technical, aligning with its stated purpose.
[EXTERNAL_DOWNLOADS]: The skill documentation references the use of several external packages, including DoubleML and econml (a Microsoft Research project). These are reputable, well-established libraries in the econometrics and data science communities and are hosted on official package registries.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process and analyze user-supplied datasets for statistical estimation. While all data ingestion represents a potential attack surface for indirect prompt injection, the skill provides legitimate analytical logic without unsafe interpolation or exploitable capabilities that would escalate risk.

causal-ml