The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the user to install the stata-mcp tool from a personal GitHub repository (https://github.com/SepineTam/stata-mcp) using the uvx tool. This introduces an external dependency from a source that is not recognized as a trusted organization or well-known service.
[DYNAMIC_EXECUTION]: The workflow generates a series of Stata .do files and executes them in sequence. This pattern of code generation followed by automated execution is a high-capability behavior that can be exploited if the inputs influencing the generation are malicious.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external paper references, datasets, and existing scripts to scaffold its analysis pipeline.
Ingestion points: Research paper pointers, datasets, and existing R scripts in the scripts/R/ directory.
Boundary markers: There are no explicit boundary markers or delimiters defined to isolate external data from the agent's instructions.
Capability inventory: The skill has access to Write, Bash, and Task tools, and it executes generated scripts via the stata-mcp server.
Sanitization: The instructions do not specify any validation or sanitization of the input data before it is interpolated into the generated executable scripts.

stata-replication